I have noticed in myself, that it helps a lot with things which are conducive to route learning - that is, it is either something which has no innate meaning, like a random sequence, or something I understand, such as a table of values.
But it helps very little (not surprisingly) when I need to learn a concept and a model. I thought maybe I can learn the model faster, if I have all the terms and their definitions down - but in fact I found it painful and probably distracting.
> But it helps very little (not surprisingly) when I need to learn a concept and a model
It helps in keeping the memory fresh. The concept you understand today may have little nuances which you will not recall if don't have some kind of revisions with some intervals.
I find it less useful for learning complex things, but more useful for retaining them. So I learn something, but then I make cards for key points so that I'll be able to remember the whole thing later. Otherwise I find that after a while I don't always recall the whole concept.
Yea, in my (limited) experience Spaced Repetition can bring larger items into memory even if the card itself is not asking a big complex question. I think if you can phrase a question in such a way that to answer it you need to draw a larger model in your head you'll gain similar benefits to more traditional micro-spaced-repetition.
Eg, you could add cards to individually remember family members, or you could add a card to remember all of the family members of X household. The latter is more vague but I think still effectively pulls the dependencies (individual family members) into your mind.
This is speculation, so take it with a grain of salt.
I'm actually working on a knowledge base / memory retention application. I'm not yet sure what I'll do for the "best UX", as I want to avoid UI work (not my area), so I expect I'll export to something like Mochi[1] or Anki.
Can you describe your format and processing a bit more? I'd like to have some understanding for potential future use cases. Eg, what are the sources (format/etc), and how would you envision them generating the Anki cards from them? Does your source currently have enough data to generate cards from? Or would someone need to go through and separate larger documents into Anki cards?
My impression is that to recall some fact, what I put in the card might not be what I care about recalling.
e.g. for Set, wikipedia has "a set is an abstract data type that can store unique values, without any particular order". Whereas what I might care about is "a Set has items, and it's quick to check if the item is in the set".
-- If I don't know what a Set is and haven't used it, it's a bit tough to come up with a description I'm going to care about.
While vocab flashcards are straightforward, non-vocab flashcards need more maintenance.
I use anki for math, computer science, languages and a lot of other things. i do this successfully (at least by my standards) and i think it works very well
Michael Nielsen (The author of the popular free online book Neural Networks and Deep Learning) has a great long form essay on using Anki titled "Augmenting Long-term memory": http://augmentingcognition.com/ltm.html
With how often I've seen spaced repetition mentioned online recently, I'm starting to assume there's a grand conspiracy that's trying to use spaced repetition to trick me into remembering about spaced repetition.
Surely if this were so, it would mean you'd see fewer mentions of spaced repetition over time, rather than more? (assuming your comment implies having seen more than usual recently)
As an aside, I love doing this sort of thing. In conversation, someone makes a joke about X doing something impossible, and then all of us who analyze everything to death start playing the scenario in our head to understand ramifications or limitations in the fictional and often impossible scenario. Drives my wife nuts, but I enjoy it so much :)
Your theory holds true with my wife and I. I love Seinfeld, she hates it :D
I should note that I actually don't find what I do comedic, nor am I doing it for comedic effect - I'm analyzing the scenario in a similar way that I find myself programming. It seems to feel.. similar to me. Eg, if we're discussing superman I might start thinking about all the normal scenarios Superman must find himself in. How he deals with those normal scenarios with abnormal abilities, and such.
Regardless, she still hates obsessing about that in the way I love haha. Similar to your statement, Seinfeld seems to take the same approach to analyzing things but for the target of comedy.
Interesting all around. Well, except to my wife. :)
I enjoy analyzing in the same way, it's just that Seinfeld takes the same tendency and turns them into jokes, which seem to be much funnier to people who enjoy breaking down trivial things.
My version of your Superman example is thinking about mutants from the X-Men universe with trivial mutations: there must be countless useless X-gene mutations, right? What might those be? Invisible skin, but not organs?
This assumes everybody started learning at the same time. Since they didn't, there have to be compromises on timing.
But just seeing the answer repeatedly isn't sufficient, you also need the effort to recall the answer. That's why so many headlines take the form of a question.
“X big website is down” just hit HN the last few weeks and people on our normally highly rational news site were convinced it was a connected conspiracy.
I'm not in school and aren't actively trying to learn one particular thing. But I've always wanted a flashcard phone app where I can browse topics, install the flashcards, and learn at my leisure. Is there a good product like that? All the apps I've seen seem to trend towards making your own cards.
While ex. Anki supports community decks (I assume others support such features, of course), part of the resistance is that making your own cards significantly contributes to learning; using premade cards is convenient, but impedes learning them.
I often feel like learning things with Anki puts memories into the wrong part of the brain. It happens way too often that I can perfectly recall something from Anki (on a PC or phone), but have a lot of trouble recalling it 'in action'.
It might be becasue you are just training recognition and not production.
I use Anki to study Russian, with great results, following Gabriel Werner's Fluent Forever.
In the book, he teaches how to learn the words and the grammar of a language using Anki. One of the fundamental ideas is to both create recognition and production cards.
The recognition cards are the ones with a word in your target language, of which you have to remember the meaning. This is what most people use SRSs for.
The production cards instead show you an image, or a sentence with blanks, and you have to come up with the correct word by yourself.
Before reading the book, I have never seen anyone using the latter, but it makes sense. You have to train your brain not only to recognize the information, but also to take it out when needed, and those are two separate paths.
What I'd really like isn't "I am studying Y using X" -I'd like "I am fluent to level Z in Y from using X."
I've used some spaced repetition/Anki stuff to put an initial dent in Portuguese. Compared to learning French the old fashioned way of practice with humans and repetition; it's a huge basket of fail for me. Virtually no useful results. Stopped with Anki, moved to Portugal; zero lessons, but making some attempts with the locals with a shitty phrasebook and dictionary -I actually got decent results. Maybe you can call that "spaced repetition." In which case everything is spaced repetition.
I believe people can remember lots of facts via flashcards, as I've done this myself in the past. I'm not so sure it works on languages.
I am just following the book and using Anki as he describes, creating cards with images from Google and sound files from Forvo.
The only advice that I find a bit rushed is to read books with the audio after you learn the top 1.000 words and half of your grammar book. That's still too hard for me.
But I can listen to this Russian-only podcast and undertand almost everything: https://russianpodcast.eu
What you describe is intentional vocabulary acquisition. However I'd like to point out that such words learned via this method consist only of a small portion of the vocabularies of both native speakers and L2 learners. Majority of vocabulary comes through incidental vocabulary acquisition [0].
Such methods of learning the language would make you a practitioner of the Skill-Building Hypothesis in your SLA (Second Language Acquisition) journey. Often which methodology to follow is decided unknowingly by the L2 learners, since the sources they utilize often dictate the methodology.
I believe it'd be my duty to inform you there's also a competing theory in the SLA field that you may fancy, called the Comprehensible Input Theory.
The book is actually also based on comprehensible input.
The learning through an SRS happens only for the 1.000 most common words of a language and its grammar.
After that, learning is intermixed with speaking, reading, writing and listening to learn vocabulary through comprehensible input.
Still, I have to say that learning the top words and grammar, I got to an impressive level, giving the effort. I can listen to some podcasts, read what people write on Facebook and some articles, and speak with friends in noisy contexts.
> It might be becasue you are just training recognition and not production.
Like trying to learn a foreign language by only reading.
There was a sheepish point in my life not too long ago where I had a bookshelf full of Spanish books that I had read, yet I couldn't follow a conversation or speak much at all.
Of course, nothing that can't be quickly fixed by dating a native-speaker that doesn't want to or can't use English. :)
I have been using Anki for the past six months for French, more-or-less following the Fluent Forever method.
I found that the method worked really well for the basic vocabulary (pictures for easily identifiable objects), cloze deletion for basic grammar and other common words. However, I have recently noticed two related problems as my deck and knowledge of the language grows is size (I have about 5000 unique cards at the moment) surrounding words with similar meanings:
- I often find that in the cloze deletion cards I insert a word that correctly completes the phrase but is not the term used on the card
- I have found that with some words I know that the word belongs in a phrase but I don't know the words specific meaning. An example using English: if the phrase was "I bit the apple" I would know that "bit" is somehow related to eating, but wouldn't remember how it differs from "eat", "chew", ect.
I'm curious if you are anyone else has had similar problems. Particularly in regards to the second problem, as it is seems to have the biggest downside. I am thinking of breaking the gold rule of the FF method and just putting definitions on the cards to make sure I actually know exactly what a word means.
The book actually allows putting definitions on cards. They just need to be in the target language, so you need a monolingual dictionary.
I have the second problem as well, but I don't know if it's really a problem. I remember learning many patterns in English before fully understanding their meaning, and the same applies to Russian.
That's probably a "feature" of learning a language. The more you will see the word in conctext, the more you will refine your understanding.
I also experience this phenomenon. In learning a language, I like to think of Anki as the primer stage in learning a word. I have to encounter and recognize that word in the wild for the memory to be fully solidified. I'll often go back to my Anki card and add a note of where I encountered it IRL, and that usually makes that card stick for good.
Yeah, me too. I try to save a new word every day that I heard in the wild. Trying to memorize random words that I didn't pick up from direct use is nearly impossible (for me, at least).
When I first tried expanding my vocabulary in a foreign language, I made the classic mistake of just logging word + definition.
I've since transitioned to rely on actual usage examples from books / in the wild. It's really the only way to actually learn, imo.
Else you'll learn that a word means "wide" but have no idea how to apply it. With usage examples, you train yourself to realize you might use this word to describe a wide road, but this other word to refer to a thick stick. And the correct word starts coming to mind in the wild, incidentally, without a deliberate step to understand the difference.
Using SRS for memorizing words is usually a mistake unless they truly are discrete pieces of information and map one to one to something you already know how to express. 99% of the time you're better off doing extensive reading. SRS works well for technical vocabulary, though.
From my podcast on SRS I linked in another comment:
> [00:11:27] These SRS flash card apps are very good for learning vocabulary and depending on what's included in the flashcards, they could be for grammar as well. But the problem for a language learner is that it's de-contextualized. So say if you just study vocabulary words, then they're going to be a lot of things that you'll miss like collocations. You won't know which words are normal to use with which other words. For example, in English, if someone asks, "How are you?" it would be completely normal to answer, "pretty good". It would also be fairly normal to answer, "absolutely fantastic". But it would be strange to answer that you're doing "absolutely good". There's no grammatical reason. It's just not something that English speakers tend to say.
> [00:12:17] And there are many, many, many language features that are like this. There are also questions of word boundaries. For example, in English, the word "nose" refers to a person's nose or a dog's nose, but not every kind of animal. For example, an elephant in English doesn't have a nose. It has a trunk. In Japanese, the same word, 鼻 (hana), is for a person's nose and the elephant's nose. So the question is, what exactly does "nose" mean? Well, really tedious language teacher could explain this for every single word that you study, or even put this on the back of every flash card for every word that you're reviewing. But, it's not going to be efficient. You'll spend so much time worrying about edge cases for every single word that you're learning that you're actually not going to get anywhere.
Extensive reading avoids all of these problems. It feels slower than cramming words at the beginning, but over time, it's a much faster way to build a functional vocabulary.
After trying to learn a few different languages the best cycle for me seems to be, remember the word in anki, then in more controlled real word setting like a class or language exchange try to use that word in conversation. If you're not willing to do that last step, I wouldn't even bother with anki if you're using it for languages
Memory in the brain is very connection based. Spaced repetition is a good technique for strengthening connections, but not for structuring it usefully.
I find the best thing is to make as many connections as possible to and amongst the things you want to learn. For vocabulary: find patterns in the spelling of the word, draw the word, use the word in a rhyme, make a joke about the word, etc. Also, connect the words together such that you can traverse a network of them in your head.
It's definitely "learned" in a very specific context. For example, I learned all US states and their capitals for a Pub Quiz. That means cards with state/capital. Then at some point there was a quiz question called "Which state capital is the last if sorted alphabetically?" And I couldn't answer it because it didn't fit my learning pattern.
I thought your comment was interesting so I posed this question to my children (10 and 12.) It took them about 15 minutes to answer. Assuming an adult might be twice as fast, that’s still a long time to answer a written quiz. So, fair point. However, they had an easier time answering other questions about extreme state capitals that had more to do with geography, so I don’t think rote memory methods should be entirely faulted.
It's better to learn things in or close to the contexts in which you will use them. That's the heart of situated learning / problem-based learning / simulations, etc:
https://en.wikipedia.org/wiki/Situated_learning
I've found that one of the best ways to learn is to combine spaced repetition with shorter, yet higher-frequency study sessions (think two daily 30-minute blocks, rather than one 1-hour block). This allows you to also take advantage of the primacy & recency effects.
According to Anki, I'm averaging 10 minutes a day on my language vocabulary deck. It does happen in a series of short, 1-2 minute bursts when I'm sitting on the toilet or whatever.
To that end, another useful thing I've discovered: Reviewing flashcards (and, spaced repetition) is not a good way to learn. It's a good way to reinforce your memory of things you've already learned. The real learning happens when you make the flashcards. The time spent focused on each concept so that you can decide how to break it into a series of smaller factoids that are right-sized for a flashcard, and coming up with whatever mnemonic devices you'll be using to help you remember this concept, is where the learning happens, because that whole process involves a fair amount of turning it over in your head.
Meaning you really shouldn't ever use a pre-made deck. It may not seem like it at first, but it really is normally more efficient to make your own from scratch.
> Meaning you really shouldn't ever use a pre-made deck.
Agreed. The process of making your own deck is too important.
Not to mention, I've never found a deck that was more than 25% applicable to me. Even "advanced" Spanish decks had a bunch of chaff and beginner-content to wade through. And you start realizing you should've just invested the time making your own starting one year ago.
I wish anki supports better import-from-text-files, i.e. I can write the cards in a text editor(e.g. vim), then import it to anki easily with syntax highlighting etc just works, I failed to find one does that, use anki's own interface to input a large amount of info is difficult for me. it will be nice if anki has a pre-defined format/template so we can edit(add/delete/change) the cards using better editors.
I'm not sure if this is exactly what you're looking for, but I've been working on a spaced repetition app[0] recently that supports markdown for cards, as well as the ability to convert markdown notes into cards.
Seems like spaced repetition works great for language learning and remembering new terms. I wonder how it could help to get better as a programmer. Any tips?
It doesn't work great for language learning, speaking a language is way more than just remembering words. Similarly, being a great programmer is not about how much syntax you can remember.
Getting better at speaking a language means speaking in it all the time. Same thing for programming - if you want to improve, your time is better spent doing actual programming rather than memorizing syntax.
I honestly wouldn't bother for programming, per se.
It's great for learning facts, so, if you wanted to, I suppose you could use it for something like memorizing a pile of library functions. But I'm having a hard time seeing that as an efficient thing to do in the age of Google and Stack Overflow and editors with auto-suggestion.
Or if you're a low-level programmer, and want to memorize a new architecture's assembly language, maybe.
Even on the language side: SRS is great for shoehorning a basic vocabulary into your head when you're first getting started with a language. It kind of sucks for learning anything but the most basic grammar, and for getting a feel for how to actually express yourself in a language, or comprehend the language as it would be spoken by a competent speaker. Comprehensible input is still the go-to method for that side of language learning. And once you get to the intermediate or high intermediate stage, SRS isn't even all that great for learning the vocabulary anymore, because it's hard to really internalize the nuances of more advanced vocabulary that way. You'll go farther faster by just watching a lot of TV and reading a lot of books, where you get to encounter the words living in their natural habitat instead of dead and dried and pinned to a notecard.
I once made anki flashcards while reading a book on linux. I think I was able to memorize a lot of useful CLI utilities because of that. It also helped me stay motivated to get through a very dry book.
Knowing what to learn is really important. The field of competitive programming probably has some good answers to your question about being a better programmer.
For anything you read that you want to remember spaced repetition is helpful. If I'm reading a book without putting questions into anki, I forget 90% of what I read. Where as with anki, I can at least keep all the basic concepts in my mind. For example, if you are reading about different types of approaches to programming, you might create a question like, "What are the aspects of Agile?"
I'm not a programmer. I work in security. Today I was on a job interview where someone asked me if I was familiar with Agile. I was able to give him the bare bones explanation of it. This is something I studied over a year ago and haven't touched since.
It can absolutely help here! I've been using Anki for about 7 years, and most of my cards (~9000) are CS-related.
I've found that focusing on the fundamentals helps for me. E.g. most programming languages have a few fundamental ideas that they're built around--memorizing those is usually good enough to quickly hop around programming languages and be productive. Algorithms, operating systems, networks, compilers, machine learning, math, etc...I grabbed some textbooks for those and just focused on the first few chapters, which lay out the fundamentals, and being able to recall and think about those in my head has helped me immensely.
Trivia is occasionally useful but it's hard to transfer to real life: e.g. even though I can answer questions about CLI command arguments when studying, I still have a hard time recalling them when sitting at a computer. Memorizing how to find the answers I need is way more useful (and a lot of this is knowing the right terminology and fundamental concepts...which are easier to recall out-of-context).
I also focus on different formulations of the same concept. Backwards and forwards retrieval: recognizing the concept from its description, and being able to describe the concept. I make separate cards for those. I also put "generative" cards for the same concept, which force me to practice using the concepts, and recognizing when to use them. And also recognizing concepts from code or other contexts. Usually just having one card for a concept isn't enough for me, unless it's really simple or is just trivia. For a hard concept, I might create 5-10 cards to help me practice recalling and using the concept in different contexts.
Very cool site! I always wondered how you were supposed to schedule Leitner boxes manually.
Somewhat related, https://kanji.koohii.com/ uses online Leitner boxes to learn how to write all the 2000+ official commonly-used Kanji in Japanese. But I believe the timing of when the cards come due is based on when they go into a box + a certain amount of time, instead of reviewing an entire box in one day (which it can do easily since it's software). It's nice to see the stacks in each (virtual) box progress over time.
I’ve used Anki pretty consistently throughout my university studies for just about every course. I plan to continue using it to keep a high level overview of some of the important maths concepts I want to stay fresh on.
I’ve been enjoying Mochi[0] as an alternative recently. It has a much nicer UI and the core features I use are all there. I find it much faster to create my cards in Markdown and with a keyboard driven workflow than what I had with Anki.
I have been patiently waiting on the Soffos[1] product from Fountech to be released to give it a try. I am assuming that there is a repetition algo in there as well.
I truly believe that spaced repetition is near-magic and it is a tragedy that it's not integrated into our society and educational system at a fundamental level. It makes me shudder to think of all the years spent learning things only to forget them weeks or months later.
In any case, I've recently embarked on a long-term educational plan utilizing spaced repetition. My goal is to learn dozens of languages, historical events, philosophical theories, lines of poetry, and various other information.
I'm curious if anyone else has used spaced repetition over a long period of time with a deliberately ambitious plan. Most people seem to just use it for dabbling in language-learning or medical school, and not as a way to embed massive amounts of knowledge in the brain.
Whenever I post about spaced repetition, people ask for my decks. You're far better off making your own decks. Read through the material you want to learn and make "notes" on what you want to remember by creating questions for yourself. By using someone else's deck, you are selling yourself short.
I like creating my own decks too. But checking out people's deck sometimes reveal interesting and creative ways to remember things.
I learnt occlusion, multiple choice questions, sound based anki from looking at other people deck.
Its like I don't like to read from other people study notes but I definitely check them out to see what I have missed.
There isn't much value in memorizing vast amounts of facts like that anymore (you already mentioned two key areas where that's still needed: language and medicine). Nowadays it's better to remember where to find facts and learn quickly than to simply remember facts upfront.
It's fine for you to have a personal goal of using spaced repetition to learn dozens of languages, historical events, etc, but it's really not a tragedy that it's not integrated into our society. It's good that we've moved on.
Sorry but I disagree completely. The ability to quickly find information online doesn’t negate the fact that intellectual progress and creativity often come from making connections between disparate fields. The only way to make these deeper connections is to truly know and digest the information, not simply to google it.
It is probably a consequence of the computer age that we think of information as a singular piece of data that is either known/nor known, and not as something that needs to be learned, dwelt upon, and digested over time.
But surely that's not true of (for example) learning one's colors in German?
Synthesis of ideas and innovation comes in areas of high ambiguity, subjectivity, and risk and no amount of spaced repetition will turn you into a Bucky Fuller or Raymond Loewy or Alfred Sloan.
There you must substitute a voracious appetite for discovery and experimentation.
Connecting information is orthogonal to remembering it. Same with digesting information: you don't need to use something like spaced-repetition to commit knowledge to long term memory in order to digest it, you just need to spend some time understanding the info. In fact, I'd argue that spaced-repetition runs somewhat counter to deeply understanding information. It's easy to repetitively memorize and recall info without actually understanding it.
I've used Pimsleur at length in the past and found it the most suitable way for me to study language. The drawbacks are the limited programs, they develop at a glacial pace, if at all anymore, and each level is very expensive. But the method is wonderful. I had friends offering to help them develop their particular language into a full course 20 years ago and they said "thanks but no thanks" and there is still nothing but a 10 lesson intro program in that language...and those intro courses are a useless waste of money except to learn how to say "I speak a little <language here>" with perfect intonation. For a long time I wished someone would unseat them but it never happened. Even related theory based sites which complimented Pimsleur like Duolingo and Memrise just got worse and worse with time.
I recommend downloading the audio from this site and creating both text and audio Anki cards for each phrase. It’s a bit like the Pimsleur process except you’ll learn 2,000+ phrases.
> I truly believe that spaced repetition is near-magic and it is a tragedy that it's not integrated into our society and educational system
Spaced repetition is absolutely key to memorizing the Qur'an [0]-- Memorizing through space-repetition is taught at almost all Madrasas (institutes of Islamic learning) [1] to students from a very young age (it isn't uncommon for students as young as 15yrs old to memorize the entire Qur'an [2]). Revisions [3] every now and then helps; and so, that is prevalent, as well.
I have been using spaced repetition for roughly 3 years and have seen some amazing personal growth.
I was born in the US, and have now learned 2 languages (Italian B2/C1, French B1/B2, and I just started Spanish but it is so similar to italian its gonna be pretty easy to pick up ).
I remember almost every book I have read in the past three years by using kindle highlights and turning them into flash- cards. I use it for Math, Computer Science, papers, new english vocabulary etc. Starting using Polar Bookshelf for IR
Some general thoughts:
1. Anki is pretty complicate, it is really important to read the manual
2. Create all cards yourself. If possible, always include sounds and images.
I'm actually looking forward to seeing what is possible in say 5-10 years, after i have read 250-700 more books and studied a few more languages. It is the single best learning hack i have ever come across.
Here is an example from The Black Swan (cloze deletion)
{{c1::The Platonic fold}} is the explosive boundary where the Platonic mind-set enters in contact with messy reality, where the gap between what you know and what you think you know becomes dangerously wide. It is here that the Black Swan is produced.
Here is a basic card I read and decide if this is info I have forgotten or if I knew about this:
“We had observed a real-estate transaction that takes place each night when we sleep. Fitting the notion of a long-wave radio signal that carries information across large geographical distances, the slow brainwaves of deep NREM had served as a courier service, transporting memory packets from a temporary storage hold (hippocampus) to a more secure, permanent home (the cortex). In doing so, sleep had helped future-proof those memories.”
Excerpt From: Matthew Walker. “Why We Sleep.” iBooks.
Did you make phonics decks to help you when you were getting started with the languages? I did that with Vietnamese and found it helped a lot in getting over the initial hump while I was visiting.
I interviewed Gabriel a couple years ago back when he was first really getting into Japanese. His pronunciation trainers are great! They're the very first Anki decks I'd reach for (or create) when starting on a new language.
It's not a cure-all, especially for language learning. I say this as someone who has done a ton of spaced repetition, language learning and who, in a former life, wrote curriculum for language learners. 95% of the time, if you have the option to choose between ER (extensive reading) and SRS, go with ER. SRS is very useful for certain sub-problems in language learning, though.
> I'm curious if anyone else has used spaced repetition over a long period of time with a deliberately ambitious plan.
An early pioneer, Piotr Wozniak, did exactly that, even to the detriment of the business success of Super Memo. He lived his life by an algorithm, timed swimming workouts to aid his SRS, put his emails through SRS, lived on a > 24 hour schedule... He basically went all in.
I built an app earlier last year to help with spaced repetition on web content: https://retaino.com
It's been interesting to see a lot of interested over the past year around the concept, though it may just be the Badder-Meinhof effect at play (which I'm now able to recall the name of thanks to Retaino!)
This is exactly what i've been looking for for a while now. signing up! have been using readwise for books, but this will be great for things beyond books! thanks!
I built splearn.io to learn faster myself through spaced repetition learning. Anyone can log in and use it 100% free though. It has offline sync and text to speech for spelling. It only uses active recall (actually tests you) for maximum efficiency.
Been using Chessable (chessable.com) for chess improvement and it really works wonders. The format is great for chess, especially for learning openings or for tactical situations where there is a clear best move.
114 comments
[ 15.2 ms ] story [ 1182 ms ] threadBut it helps very little (not surprisingly) when I need to learn a concept and a model. I thought maybe I can learn the model faster, if I have all the terms and their definitions down - but in fact I found it painful and probably distracting.
It helps in keeping the memory fresh. The concept you understand today may have little nuances which you will not recall if don't have some kind of revisions with some intervals.
Eg, you could add cards to individually remember family members, or you could add a card to remember all of the family members of X household. The latter is more vague but I think still effectively pulls the dependencies (individual family members) into your mind.
This is speculation, so take it with a grain of salt.
I try to add new words/concepts/things etc. I learn into Anki.
IMO, it greatly helps me with memorising various and more obscure terms, that I used to always forget.
Can you describe your format and processing a bit more? I'd like to have some understanding for potential future use cases. Eg, what are the sources (format/etc), and how would you envision them generating the Anki cards from them? Does your source currently have enough data to generate cards from? Or would someone need to go through and separate larger documents into Anki cards?
Appreciate any detail you can give :)
[1]: https://mochi.cards
https://www.primerlabs.io
Primer generates Anki apkg files too.
I am way behind schedule to launch it. But I am working as hard as I can to ship soon.
My impression is that to recall some fact, what I put in the card might not be what I care about recalling. e.g. for Set, wikipedia has "a set is an abstract data type that can store unique values, without any particular order". Whereas what I might care about is "a Set has items, and it's quick to check if the item is in the set". -- If I don't know what a Set is and haven't used it, it's a bit tough to come up with a description I'm going to care about.
While vocab flashcards are straightforward, non-vocab flashcards need more maintenance.
Surely if this were so, it would mean you'd see fewer mentions of spaced repetition over time, rather than more? (assuming your comment implies having seen more than usual recently)
As an aside, I love doing this sort of thing. In conversation, someone makes a joke about X doing something impossible, and then all of us who analyze everything to death start playing the scenario in our head to understand ramifications or limitations in the fictional and often impossible scenario. Drives my wife nuts, but I enjoy it so much :)
Tell your wife you're using a proven comedic technique.
"You see, I'm actually being funny right now. It's comedy, trust me."
I should note that I actually don't find what I do comedic, nor am I doing it for comedic effect - I'm analyzing the scenario in a similar way that I find myself programming. It seems to feel.. similar to me. Eg, if we're discussing superman I might start thinking about all the normal scenarios Superman must find himself in. How he deals with those normal scenarios with abnormal abilities, and such.
Regardless, she still hates obsessing about that in the way I love haha. Similar to your statement, Seinfeld seems to take the same approach to analyzing things but for the target of comedy.
Interesting all around. Well, except to my wife. :)
My version of your Superman example is thinking about mutants from the X-Men universe with trivial mutations: there must be countless useless X-gene mutations, right? What might those be? Invisible skin, but not organs?
But just seeing the answer repeatedly isn't sufficient, you also need the effort to recall the answer. That's why so many headlines take the form of a question.
Only if they start learning the meaning behind it by using it ;) As long as they ignore it, it will just reappear in a fixed interval.
The fact that spaced repetition is proving valuable now is because people in general are confused about their motives for learning stuff.
Anyhow, when it comes to learning, motivation is primary, not method.
[0]:https://en.wikipedia.org/wiki/Baader%E2%80%93Meinhof_effect
(I think it's called Memory Palace.)
I use Anki to study Russian, with great results, following Gabriel Werner's Fluent Forever.
In the book, he teaches how to learn the words and the grammar of a language using Anki. One of the fundamental ideas is to both create recognition and production cards.
The recognition cards are the ones with a word in your target language, of which you have to remember the meaning. This is what most people use SRSs for.
The production cards instead show you an image, or a sentence with blanks, and you have to come up with the correct word by yourself.
Before reading the book, I have never seen anyone using the latter, but it makes sense. You have to train your brain not only to recognize the information, but also to take it out when needed, and those are two separate paths.
[Update: fixed the book's author name]
I've used some spaced repetition/Anki stuff to put an initial dent in Portuguese. Compared to learning French the old fashioned way of practice with humans and repetition; it's a huge basket of fail for me. Virtually no useful results. Stopped with Anki, moved to Portugal; zero lessons, but making some attempts with the locals with a shitty phrasebook and dictionary -I actually got decent results. Maybe you can call that "spaced repetition." In which case everything is spaced repetition.
I believe people can remember lots of facts via flashcards, as I've done this myself in the past. I'm not so sure it works on languages.
The only advice that I find a bit rushed is to read books with the audio after you learn the top 1.000 words and half of your grammar book. That's still too hard for me.
But I can listen to this Russian-only podcast and undertand almost everything: https://russianpodcast.eu
Such methods of learning the language would make you a practitioner of the Skill-Building Hypothesis in your SLA (Second Language Acquisition) journey. Often which methodology to follow is decided unknowingly by the L2 learners, since the sources they utilize often dictate the methodology.
I believe it'd be my duty to inform you there's also a competing theory in the SLA field that you may fancy, called the Comprehensible Input Theory.
[0]: https://sci-hub.tw/10.2307/747758
The learning through an SRS happens only for the 1.000 most common words of a language and its grammar.
After that, learning is intermixed with speaking, reading, writing and listening to learn vocabulary through comprehensible input.
Still, I have to say that learning the top words and grammar, I got to an impressive level, giving the effort. I can listen to some podcasts, read what people write on Facebook and some articles, and speak with friends in noisy contexts.
Like trying to learn a foreign language by only reading.
There was a sheepish point in my life not too long ago where I had a bookshelf full of Spanish books that I had read, yet I couldn't follow a conversation or speak much at all.
Of course, nothing that can't be quickly fixed by dating a native-speaker that doesn't want to or can't use English. :)
I found that the method worked really well for the basic vocabulary (pictures for easily identifiable objects), cloze deletion for basic grammar and other common words. However, I have recently noticed two related problems as my deck and knowledge of the language grows is size (I have about 5000 unique cards at the moment) surrounding words with similar meanings:
- I often find that in the cloze deletion cards I insert a word that correctly completes the phrase but is not the term used on the card
- I have found that with some words I know that the word belongs in a phrase but I don't know the words specific meaning. An example using English: if the phrase was "I bit the apple" I would know that "bit" is somehow related to eating, but wouldn't remember how it differs from "eat", "chew", ect.
I'm curious if you are anyone else has had similar problems. Particularly in regards to the second problem, as it is seems to have the biggest downside. I am thinking of breaking the gold rule of the FF method and just putting definitions on the cards to make sure I actually know exactly what a word means.
I have the second problem as well, but I don't know if it's really a problem. I remember learning many patterns in English before fully understanding their meaning, and the same applies to Russian.
That's probably a "feature" of learning a language. The more you will see the word in conctext, the more you will refine your understanding.
I've since transitioned to rely on actual usage examples from books / in the wild. It's really the only way to actually learn, imo.
Else you'll learn that a word means "wide" but have no idea how to apply it. With usage examples, you train yourself to realize you might use this word to describe a wide road, but this other word to refer to a thick stick. And the correct word starts coming to mind in the wild, incidentally, without a deliberate step to understand the difference.
Definition alone is basically useless.
From my podcast on SRS I linked in another comment:
> [00:11:27] These SRS flash card apps are very good for learning vocabulary and depending on what's included in the flashcards, they could be for grammar as well. But the problem for a language learner is that it's de-contextualized. So say if you just study vocabulary words, then they're going to be a lot of things that you'll miss like collocations. You won't know which words are normal to use with which other words. For example, in English, if someone asks, "How are you?" it would be completely normal to answer, "pretty good". It would also be fairly normal to answer, "absolutely fantastic". But it would be strange to answer that you're doing "absolutely good". There's no grammatical reason. It's just not something that English speakers tend to say.
> [00:12:17] And there are many, many, many language features that are like this. There are also questions of word boundaries. For example, in English, the word "nose" refers to a person's nose or a dog's nose, but not every kind of animal. For example, an elephant in English doesn't have a nose. It has a trunk. In Japanese, the same word, 鼻 (hana), is for a person's nose and the elephant's nose. So the question is, what exactly does "nose" mean? Well, really tedious language teacher could explain this for every single word that you study, or even put this on the back of every flash card for every word that you're reviewing. But, it's not going to be efficient. You'll spend so much time worrying about edge cases for every single word that you're learning that you're actually not going to get anywhere.
Extensive reading avoids all of these problems. It feels slower than cramming words at the beginning, but over time, it's a much faster way to build a functional vocabulary.
I find the best thing is to make as many connections as possible to and amongst the things you want to learn. For vocabulary: find patterns in the spelling of the word, draw the word, use the word in a rhyme, make a joke about the word, etc. Also, connect the words together such that you can traverse a network of them in your head.
It's better to learn things in or close to the contexts in which you will use them. That's the heart of situated learning / problem-based learning / simulations, etc: https://en.wikipedia.org/wiki/Situated_learning
According to Anki, I'm averaging 10 minutes a day on my language vocabulary deck. It does happen in a series of short, 1-2 minute bursts when I'm sitting on the toilet or whatever.
To that end, another useful thing I've discovered: Reviewing flashcards (and, spaced repetition) is not a good way to learn. It's a good way to reinforce your memory of things you've already learned. The real learning happens when you make the flashcards. The time spent focused on each concept so that you can decide how to break it into a series of smaller factoids that are right-sized for a flashcard, and coming up with whatever mnemonic devices you'll be using to help you remember this concept, is where the learning happens, because that whole process involves a fair amount of turning it over in your head.
Meaning you really shouldn't ever use a pre-made deck. It may not seem like it at first, but it really is normally more efficient to make your own from scratch.
Agreed. The process of making your own deck is too important.
Not to mention, I've never found a deck that was more than 25% applicable to me. Even "advanced" Spanish decks had a bunch of chaff and beginner-content to wade through. And you start realizing you should've just invested the time making your own starting one year ago.
[0] https://mochi.cards
Getting better at speaking a language means speaking in it all the time. Same thing for programming - if you want to improve, your time is better spent doing actual programming rather than memorizing syntax.
It's great for learning facts, so, if you wanted to, I suppose you could use it for something like memorizing a pile of library functions. But I'm having a hard time seeing that as an efficient thing to do in the age of Google and Stack Overflow and editors with auto-suggestion.
Or if you're a low-level programmer, and want to memorize a new architecture's assembly language, maybe.
Even on the language side: SRS is great for shoehorning a basic vocabulary into your head when you're first getting started with a language. It kind of sucks for learning anything but the most basic grammar, and for getting a feel for how to actually express yourself in a language, or comprehend the language as it would be spoken by a competent speaker. Comprehensible input is still the go-to method for that side of language learning. And once you get to the intermediate or high intermediate stage, SRS isn't even all that great for learning the vocabulary anymore, because it's hard to really internalize the nuances of more advanced vocabulary that way. You'll go farther faster by just watching a lot of TV and reading a lot of books, where you get to encounter the words living in their natural habitat instead of dead and dried and pinned to a notecard.
I'm not a programmer. I work in security. Today I was on a job interview where someone asked me if I was familiar with Agile. I was able to give him the bare bones explanation of it. This is something I studied over a year ago and haven't touched since.
I've found that focusing on the fundamentals helps for me. E.g. most programming languages have a few fundamental ideas that they're built around--memorizing those is usually good enough to quickly hop around programming languages and be productive. Algorithms, operating systems, networks, compilers, machine learning, math, etc...I grabbed some textbooks for those and just focused on the first few chapters, which lay out the fundamentals, and being able to recall and think about those in my head has helped me immensely.
Trivia is occasionally useful but it's hard to transfer to real life: e.g. even though I can answer questions about CLI command arguments when studying, I still have a hard time recalling them when sitting at a computer. Memorizing how to find the answers I need is way more useful (and a lot of this is knowing the right terminology and fundamental concepts...which are easier to recall out-of-context).
I also focus on different formulations of the same concept. Backwards and forwards retrieval: recognizing the concept from its description, and being able to describe the concept. I make separate cards for those. I also put "generative" cards for the same concept, which force me to practice using the concepts, and recognizing when to use them. And also recognizing concepts from code or other contexts. Usually just having one card for a concept isn't enough for me, unless it's really simple or is just trivia. For a hard concept, I might create 5-10 cards to help me practice recalling and using the concept in different contexts.
Somewhat related, https://kanji.koohii.com/ uses online Leitner boxes to learn how to write all the 2000+ official commonly-used Kanji in Japanese. But I believe the timing of when the cards come due is based on when they go into a box + a certain amount of time, instead of reviewing an entire box in one day (which it can do easily since it's software). It's nice to see the stacks in each (virtual) box progress over time.
I’ve been enjoying Mochi[0] as an alternative recently. It has a much nicer UI and the core features I use are all there. I find it much faster to create my cards in Markdown and with a keyboard driven workflow than what I had with Anki.
0. https://mochi.cards/
I can see how using markdown to create cards would be a big draw, though.
[1] https://www.fountech.ai/work
In any case, I've recently embarked on a long-term educational plan utilizing spaced repetition. My goal is to learn dozens of languages, historical events, philosophical theories, lines of poetry, and various other information.
I'm curious if anyone else has used spaced repetition over a long period of time with a deliberately ambitious plan. Most people seem to just use it for dabbling in language-learning or medical school, and not as a way to embed massive amounts of knowledge in the brain.
Its like I don't like to read from other people study notes but I definitely check them out to see what I have missed.
It's fine for you to have a personal goal of using spaced repetition to learn dozens of languages, historical events, etc, but it's really not a tragedy that it's not integrated into our society. It's good that we've moved on.
It is probably a consequence of the computer age that we think of information as a singular piece of data that is either known/nor known, and not as something that needs to be learned, dwelt upon, and digested over time.
Synthesis of ideas and innovation comes in areas of high ambiguity, subjectivity, and risk and no amount of spaced repetition will turn you into a Bucky Fuller or Raymond Loewy or Alfred Sloan.
There you must substitute a voracious appetite for discovery and experimentation.
https://www.goethe-verlag.com/
Spaced repetition is absolutely key to memorizing the Qur'an [0]-- Memorizing through space-repetition is taught at almost all Madrasas (institutes of Islamic learning) [1] to students from a very young age (it isn't uncommon for students as young as 15yrs old to memorize the entire Qur'an [2]). Revisions [3] every now and then helps; and so, that is prevalent, as well.
[0] https://sfjamaat.org/sf/mtb/hifz/A%20practical%20guide%20for...
[1] https://en.wikipedia.org/wiki/Madrasa
[2] Two key orthogonal parts to memorizing the Qur'an are the rules of recitation (Tajwid) and reading (Qira'at), which prove difficult for non-Arabs.
[2] https://en.wikipedia.org/wiki/Test_preparation
I was born in the US, and have now learned 2 languages (Italian B2/C1, French B1/B2, and I just started Spanish but it is so similar to italian its gonna be pretty easy to pick up ).
I remember almost every book I have read in the past three years by using kindle highlights and turning them into flash- cards. I use it for Math, Computer Science, papers, new english vocabulary etc. Starting using Polar Bookshelf for IR
Some general thoughts:
1. Anki is pretty complicate, it is really important to read the manual
2. Create all cards yourself. If possible, always include sounds and images.
3. Follow these steps when creating cards: https://www.supermemo.com/en/archives1990-2015/articles/20ru...
4. Do your reviews every day
I'm actually looking forward to seeing what is possible in say 5-10 years, after i have read 250-700 more books and studied a few more languages. It is the single best learning hack i have ever come across.
{{c1::The Platonic fold}} is the explosive boundary where the Platonic mind-set enters in contact with messy reality, where the gap between what you know and what you think you know becomes dangerously wide. It is here that the Black Swan is produced.
Here is a basic card I read and decide if this is info I have forgotten or if I knew about this:
“We had observed a real-estate transaction that takes place each night when we sleep. Fitting the notion of a long-wave radio signal that carries information across large geographical distances, the slow brainwaves of deep NREM had served as a courier service, transporting memory packets from a temporary storage hold (hippocampus) to a more secure, permanent home (the cortex). In doing so, sleep had helped future-proof those memories.”
Excerpt From: Matthew Walker. “Why We Sleep.” iBooks.
> I'm curious if anyone else has used spaced repetition over a long period of time with a deliberately ambitious plan.
An early pioneer, Piotr Wozniak, did exactly that, even to the detriment of the business success of Super Memo. He lived his life by an algorithm, timed swimming workouts to aid his SRS, put his emails through SRS, lived on a > 24 hour schedule... He basically went all in.
If you're interested in SRS and its history, you might enjoy my podcast: https://alchemist.camp/learning-machine/spaced-repetition-sy...
It's been interesting to see a lot of interested over the past year around the concept, though it may just be the Badder-Meinhof effect at play (which I'm now able to recall the name of thanks to Retaino!)
I recently created an SRS app specific to learning Chinese vocabulary. You can check it out at:
https://www.dailychinese.app
[1] https://www.penguinrandomhouse.com/books/221559/how-we-learn...
Other threads: https://hn.algolia.com/?query=Spaced%20Repetition%20comments...