Hacker News new | past | comments | ask | show | jobs | submit login
Spaced repetition (gwern.net)
299 points by ergot on Dec 11, 2016 | hide | past | favorite | 115 comments

I find it very frustrating that spaced repetition is the best recourse we (humans) have to remember information. I am regularly concerned with the number of thoughts and interesting ideas I have each day, and don't manage to write down or otherwise preserve. I have begun recording some minor annotations and idle thoughts on paper a month ago and after reviewing them today, I am shocked at how many I simply do not recall writing down at all. Worse, some of the information has lost context: I wrote it in a certain state of mind, certain emotional state which I no longer have and so the preserved words on paper are missing meaning and appear alien. The situation is so poor that it seems 20% of the material I have recorded has not even been properly preserved from only a month ago.

I find memory endlessly frustrating; we have so many unique thoughts and insights and we are forced to watch them circle the drain. Worse yet, most people aren't even aware this is happening, no internal inventory has been taken to show the poor state of affairs!

The tech community has a fixation on life extension, but I am worried I am not even properly remembering the life I am living right now, nevermind tomorrow.

I live and work in the country, by a stream. I spend endless hours watching the stream. It is absolutely filled with wildlife; from birds, to turtles, to fish. An interesting thing about watching a stream is that, at first you may be unaware that there are any fish in it at all. I find that it takes me about 2-3 minutes before I spot the first one. Then after time I see another, and another and another. Eventually I am aware that there are hundreds, if not thousands of fish of all shapes and sizes swimming just under the surface. It is incredible, because only moments before I will have seen none of them.

The interesting thing about the stream is that it is different each day. The stream I live near is a tidal stream, since I live by the ocean. Sometimes the tide is high when I go, so the water is brackish. At those times, I can see puffer fish and some kind of long narrow fish with large eyes and fins that look a bit like wings. When the tide is low, the carp swim from upstream along with the turtles. Blue herons stand in the shallower sections trying to catch the endless minnows that skip through the pebbles in the river bed.

I have often thought to record the stream by taking pictures, or video. But then, to what end? What time shall I go? And in what season? And, if I diligently point out the numerous animals that fill the stream, will I not deprive the viewer of the greatest pleasure -- that of discovering an amazing world where one would think that nothing existed?

I think your stream of consciousness is similar. The value is not in the individual thoughts and ideas, but in the experience of those ideas rushing from one place to another. While it is important to spend time to reflect on and view this wonder, it is not necessary to preserve a snapshot of it. Doing so will only distract you from what is happening now.

A person's life is too short to do all the things he or she imagines even in a single day. Do not waste your lifetime by chasing few ephemeral ideas.

Your analogy is pretty and while I agree with the sentiment I disagree with the scenario presented. I am glad I don't remember everything, yes, clearly perfect memory would be a burden. Rather I am frustrated by my lack of violation, I am not making a conscious choice to remember or forget; the decision is made for me by my flawed existence. The inability to exercise my will and make a meaningful decision over what I remember frustrates me.

That you think life is short is merely because you've forgotten most of it. When you stare at an analog clock, the seconds feel like long eternity.

> Rather I am frustrated by my lack of violation, I am not making a conscious choice to remember or forget; the decision is made for me by my flawed existence. The inability to exercise my will and make a meaningful decision over what I remember frustrates me.

You think your existence is flawed. Have you concsiously choosen this? You just happened to like the idea that you can be better than you are when you first heard of this. And after multiple tries and fails it feels/looks like something is flawed. But I think it's just a downside of that idea. Your observations clearly show that you cannot control your mind. It's just that sometimes visions of better me are in line with what's happening and sometime's are not.

Certainly brings to mind Siddhartha and what he learned from watching the river.

"But of all the water's secrets, he saw today only a single one-one that struck his soul. He saw that this water flowed and flowed, it was constantly flowing, and yet it was always there; it was always eternally the same and yet new at every moment!"

Well, some ideas make people rich, so I don't know about that last part, but I like your analogy.

I find it very frustrating that spaced repetition is the best recourse we (humans) have to remember information.

I've worked on this for a while. I've written several different memory programs, for helping to review and memorize things. I have some thoughts that may help you feel better:

In the end, spaced repetition is not what causes people to memorize things. You can review an object over and over and over......50 times, or a hundred times, and still not remember it. Other times, something will click in your mind, and you'll remember it the first time, without any repetition.

My hypothesis is that repetition merely increases the chances of your brain being in a state where it will remember the thing you are trying to remember. And certainly there are things you can do to increase the chances of it sticking.

Having a Memory is only half the equation recall is the other. Take a movie you have not watched in a decade and write down what you recall about it and how it ends. Then watch the first half and write down what you remember about the second half. Chances are good you will be able to add a lot more details the second time.

Spaced repition aids both parts of this so not only is the information stored it's also easy to retrieve.

I think this is a big part of the trick. To remember something you need to practice recalling it.

Spaced repetition software is designed to maximise absolute throughput. That is, maximise the quantity items-remembered/year.

The way it does this is to open a firehose at your memory. If the information is simple and well presented, then ideally about 90-95% of it will stick.

If you want to retain 1000s of items (like a foreign language) and can accept some amount of loss, then spaced repetition is effective.

But it's not designed to increase the relative chance that any individual item will be remembered. If you want to learn a small number of items with 100% recall, then spaced repetition is a bad way of doing that.


But it's not designed to increase the relative chance that any individual item will be remembered.

You're talking about what it was designed to do, I'm hypothesizing about what it's actually doing.

> In the end, spaced repetition is not what causes people to memorize things. You can review an object over and over and over......50 times, or a hundred times, and still not remember it.

Ebbinghaus demonstrated that spaced repetition is what causes people to memorize things, using nonsense words for exactly that purpose, to ensure that there was no possibility of any sort of 'clicking' or semantic processes. Just pure memory based on repetition either massed or spaced.

what is connected sticks. you remember one thing because it connects to something else you know and you forget another as it's not connected. or much more likely: it is actually in your memory (ie neutral pathways that 'represent' that item exist, but you have no way of actively or consciously recalling it as you cannot find those pathways in your own mind. all that works by association, so connection is key - and then you need strong connection and repeated use to make sure it's not overwritten (the connections get weaker as each neuron makes new or uses other connections) and continues to be connected.

People who really get into the idea of maintaining a large personal information store generally seem to start fashioning technology that suits them, anything from paper, cards, and filing boxes to original software development.

And the problem of information loss is, indeed, well known. Anyone who has to take meeting minutes and distil them into a report faces a huge task; human conversation regularly drifts off topic.

We have such a highly connected world now that in-brain memory, while still undeniably valuable, isn't quite as important as it used to be - certainly not as much as it was in the ancient era, where oral traditions were the rule. If you want to know general facts and principles, these days, you hit up a search engine. The only part that has to be committed to memory now is the practice and understanding necessary for subject fluency. I don't concern myself with knowing everything, but with being able to act naturally moment-to-moment, following small whims, and yet still progress towards some greater task with a few habits and techniques that will develop it regardless of the angle I am using at that moment.

For my own ideas I use a variety of techniques now: To quickly develop stuff in the near term, I open an empty, private chat room and monologue into it. To visualize, brainstorm, and recall larger things, I make mind maps. For basic to-dos I use plain text or Trello. When I need visuals, I keep around some binder paper to doodle with.

I can completely relate to that frustration. I came from an engineering background before I went to medical school and I struggled a lot with the different styles of learning (applied principles at MIT vs. knowledge acquisition in med school). The biggest mistake you can make is think that your ability to remember is just an innate ability instead of a skill that can be cultivated.

The problem with spaced repetition software is that "SR" is just one of many known cognitive principles to aid in learning and memory. The encoding process is just as, if not more important than, the review process. I highly encourage looking into interleaving vs blocking, overlearning, learning vs. performance, retrieval-induced forgetting, contextual variability, and the (pre)testing effect. Learning how to incorporate these into your existing study process could really help you out.

This is where great existing software (e.g. Anki) falls flat for many people. If you're interested in seeing an alternative approach you can check out my company's platform Memorang (https://www.memorangapp.com). We recently raised a $500K seed round and are working on some really interesting problems around learning and memory, including a new type of spaced repetition that deviates significantly from the deterministic algorithms like supermemo3/Anki. Probably even more interesting, in addition to building apps for students we're also implementing a platform for developers and education entrepreneurs to monetize their efforts.

Although we're still fairly new we finished second to duolingo at the international reimagine education awards last week. Check out Memorang (web/Android) and let us know what you think (or come join us as a dev --> founders@memorangapp.com).

Wish it could import Supermemo knowledge (with history). Simply can't afford switching to a new tool at this point (7 year Supermemo user with thousands of knowledge items).

If i may ask: how do you feel supermemo aided you? How regularly/much do you use it? What items do you store and what card models have you come to prefer (cloze, q-a, ...?) How has this impacted your life?

I have seen some kind of supermemo import for anki, not sure if it's maintained. Possibly it was just a script someone created - sorry can't find it right now.

I use it to memorize Latin (and some Greek) vocabulary. What I use mostly is Q-A. Typically it looks like this:

Q. lat. adj. black (shiny)

A. niger, -a, -um

And the other way:

Q. niger, -a, -um

A. black (shiny)

So when I enter the knowledge, I write in a .txt file the first form, and the script I run generates the second. (I read on Supermemo's website it's good to test the reverse case too, so by default I do this for all knowledge items). Over time I found it was useful to prefix vocabulary with the language (lat.) because you'll typically have the same words in another language (grk.) and if it's an adjective or adverb to append (adj./adv.) after the language prefix (I wanted a single database). For verbs I found it's sufficient to write the English part as the infinite to .. (e.g. Q. lat. to kill, A. neco (1). I will occasionally write the word with macrons in brackets after the word if it's weird, e.g.

Q. lat. adv. daily

A. cotidie (cotIdiE)

I wanted to separate the two because real Latin literature does not contain the long/short macros and I wanted to condition myself to that from the start.

I use cloze mainly for grammar.

I use Supermemo daily and it takes about 20-30 minutes to memorize what is currently ~6500 items. My daily workload is between 70-100 items and my recall is about 85%. I will admit that I'm extremely hard on myself and will only score a 5 if it's absolutely perfect.

Whatever I've looked at before of Supermemo -> Anki import did not get it right 100%. I don't want my repetition history or workload screwed around with. I also want the ability to go back to Supermemo if it doesn't work out. Unless I feel comfortable that I can switch between both systems seamlessly I'll probably never switch. I run my Supermemo on a Windows VirtualBox in Linux so it can stay that way forever.

I would probably never use Supermemo for a "live" language. It would be a waste of time, imho. For languages which are not really spoken like Attic Greek or Latin, it is indispensable.

My only real problem with Supermemo is that there is no way to access it over the Internet. The database lives in that Virtualbox Windows. It would be amazing if it had the ability to spawn a little webserver that you could do your drills from.

> a new type of spaced repetition that deviates significantly from the deterministic algorithms

Could you go into more detail about this?

And then what?

You remember everything about your life - for what benefit/purpose/reason? Even if we could remember everything we don't have the time, attention, or brainpower to dutifully filter and process three hundred million seconds (~10-20 years) of memories before making a decision, every decision.

You write down all your interesting thoughts and ideas - for who, for what? Other people have already a lifetime of thoughts themselves and more they wish they could think.

Is it almost more interesting to forget, and then you can re-experience things anew and not get bored of 'seen that, done that, been there'?

I wonder if one meaningful way forward is artificial brains/compute engines (probably required to be sentient AI - but maybe not?) which can combine and work on the thoughts and scale of thousands, then millions, of people.

"And then what?"

Nutritional deficiencies. Aging. Lack of sleep. A variety of diseases. Use of certain drugs (like marijuana).

All of these have been documented to affect memory.

Also, I think one's attitudes towards the past will affect one's memory of it. For example, I am very present- and future-oriented. I rarely dwell on the past or review my day (or my past life in general), I don't keep a journal. I also don't ascribe much importance to most day to day events. As a result (combined with some of the other factors I listed above), I don't really have a great long-term memory. Many of the things in my childhood, and even up to college and beyond are lost to me. Even when I've kept a journal, when I've gone back, I didn't remember many of the things in it. However, I think if I reviewed my journal from time to time, thought about what happened in the past more, or made an effort to remember it, I'd have a better long-term memry. Spaced repetition.

No, I meant more "why do you (parent poster) want to remember everything?". OK so hypothetical world where you remember everything ... then what? What's different, why do you want that? Humans don't have the brainpower to make use of memories of everything.

'Having loads of memories' is not exactly the same thing as 'learning more' or 'having clear focus on a goal', or 'intuition', or 'improved decision making', or 'being able to do more things', or 'knowledge', or 'wisdom' - and "I wish I remembered everything" sounds more on the lines of anxiety or hoarding, than a genuinely useful thing to desire.

And that wasn't said in the context of learning a subject, it was in the context of "remembering moment by moment thoughts" and "living well". Specifically, how do you live more-well by remembering what you thought about music one day four years ago when watering some house plants?

How many minor things do you think you've made a mistake over, repeatedly, but never learned a lesson from because each individual time was separated by months so you'd forgotten by the time you do it next?

The answer is, you don't know, because you can't remember any of them.

If they weren't bad enough mistakes to remember, how much can they really matter?

There are techniques with better payoff and retention. Check the memory systems advocated by David Roth or Harry Lorayne, or the locus method (https://en.m.wikipedia.org/wiki/Method_of_loci)

Read also: Moonwalking with Einstein

Also see the Mentat Wiki:


They've got many great, practical articles on mnemonics and other memorization techniques.

Seconding this recommendation -- this book was a really fun (and fast!) read and was really eye-opening about how real memory tricks work and started to get me thinking about the utility of having vast amounts of information memorized.

How do you know they have better payoff and retention?

I don't have any studies, but check the wikipedia page for examples of Memory Championship winners using the method to, for example, memorize a deck of shuffled cards or a random sequence of digits. I bet Moonwalking with Einstein would have some references in the scientific literature.

The issue with the methods is that you have to use them actively. As the author's mentor says in the book: most of the memory champions have horrible memory in normal situations, they were just desperate to change that, so they pursued mnemonic techniques. You might have an elaborate system to remember hundreds of playing cards in a loci palace (which takes months and lots of repetition to develop!!) and still frequently lose your key or forget names.

You can develop/discover and consciously apply mnemonic techniques for each situation (eg remembering names at a party by associating the name with prominent features of the face or funny analogies) but if you don't use them they'll be of little help.

Bill Clinton (or was it JFK??) is said to have all his life kept a notebook with him and scribbled down any personal facts revealed in a situation - name, alma mater, etc - and was then able to recall them when meeting the same person years later. He must have had an impressive system and much dedication - but also must have been a bit awkward to talk to when everything you say is recorded (apparently a good listener though).

Also http://artofmemory.com/ for 5,000+ pages of free content about memorization techniques.

Are you taking any anti-anxiety medications like Clonazepam? Many medications have a negative effect on memory, but benzodiazepines are well known for this.

First thing to check when you suspect memory issues are medications.

I'm on no prescriptions of any kind. I am not describing memory loss, but instead the amount of thoughts lost in "filler time". What were you thinking about between the parking lot and the front door? What did you think about while brushing your teeth? How did you decide what to eat today? A lot of mental decision making and "down time" thoughts never end up in short or long term memory at all.

Mindfulness meditation as described by Jon Kabat-Zinn might be helpful to you. I would describe it as learning to become more aware of yourself, including becoming more aware of your own thinking.

Most of us, even if we know that we have filler time (I call it autopilot), aren't aware of how often it occurs.

Be careful though, some of that filler time is a necessary part of memory consolidation and baking in new things we have learned. Eliminating it entirely would probably be counterproductive.

Just my speculation of course.

Not sure if spaced repetition is the best resource we have. I haven't researched it, but the question I have is: how much of everyday's information is stored in the unconscious. Using hypnosis people can remember long forgotten things(assuming these are actual memories and not produced by imagination).

I have actively tried to use unconscious processes to remember some trivial stuff. Easiest way to experiment this is when you can't remember someone's name. You know the feeling when you almost got the name, but it still manages to escape your mind. What I have done is just say to myself: This name will pop into my mind once I stop trying to remember it. And more often than not, once I have concentrated on something else(it usually takes few minutes), this name will pop suddenly into my mind. Same technique can be used with lost keys etc.

I find it interesting that we can schedule some tasks in our minds. A la "Start searching my lost keys, please". And upon completion it will trigger a notification.

What also helps is to try to actively recall the context in which an item was learned: What's her name? Oh, i met her at a party. It was at Steve's place. She was holding a glass, possibly with water. She was with someone, i think her cousin... etc

I deal with this by adding as much state-of-mind detail to my notes and idea-scribblings as I feel necessary to enable me to come back and maybe use it later. I don't worry about why I can't remember my exact state of mind, because even in operating systems, process-checkpointing and -migration is difficult. Front-load that difficulty with metadata!

I'll just throw this here

I've tried to make a point of commenting more and more on news articles and discussion boards

not because the world really needs to hear my thoughts

but it seems like a useful way of capturing my still fairly rough thoughts about whatever I just read (with the nice bonus that sometimes the rest of the universe comes along and tells you what's wrong with your thought, or how to improve your thoughts)

its a way to ever so slightly embed my ideas in my head

(having typed that out, the thought occurs to me that maybe I should occasionally make anki flashcards out of comments that seem especially worth remembering)

------- -------

writing stuff down, it used to bother me that I'd have ideas, and then by the time I tried to turn them into writing, the elegance of the idea had seemed to evaporate

or that the mechanical process of writing it down had distracted me from improving my idea

I've come to think that that's not really good way to think about it

that if I can't satisfactorily write down my idea, it needed improving anyways

Exercise makes you learn better: http://fitness.mercola.com/sites/fitness/archive/2012/10/12/...

Walk every day, reviewing your interesting ideas, they will last with you.

I did and continue trying to do (it is harder now).

There's a very convincing and well researched book on this issue: Spark! How exercise will improve the performance of your brain.

The gist: aerobic exercise helps neurogenesis and connectivity in the brain. Running is best if i remember correctly, or maybe that was just my private conclusion.

> I find it very frustrating that spaced repetition is the best recourse we (humans) have to remember information.

It's not. At least, not at the expense of other things. Memorization is but one piece.

The book "Make it Stick" goes into this and a lot of other techniques that blend together, but SRS is not the whole of it by a long ways.

Would you mind telling me what those other techniques are?

Have you used them, and how well have they worked for you?

I'm trying, and I have no basis to back this up, so value this what you are paying for it... =)

A couple of the techniques are:

* Mix things up. Doing "learning" in an all of one thing, then all of another, is less helpful than mixing up topics. Baseball players that practice hitting 20 fastballs, then 20 sliders, etc. do less well OVERALL than one where the pitcher randomly picks a type of pitch one after another.

* Learn in different ways, or using different answers. If you're learning to sight read music, read the measures backwards. Or down one side and up the other. When you're memorizing words, come up with a different sentence to use the word in, each time. Come up with a word that rhymes with the correct word. Describe a picture that would characterize the word/concept. Sing the word in a song.

* Synthesize, synthesize, synthesize. (overlapping the previous point) - when coming up with an answer to a question, consider different questions that have the same answer. What things are LIKE this answer? What things are different? Picture yourself using this word/concept in some setting. (Things I like to do are picturing myself in my house. Or outside. Or in the winter, or naked, or with a famous person (perhaps also naked).) The point is to create imagery associated with the thing you're learning.

A lot of this boils down to getting at the memory through many different neural pathways. SRS is just a way to strengthen ONE of those ways, which is important, but ideally you want to use many. One strong rope holding you from falling down a cliff is good. Many, possibly even weaker ones, can be much better.

exactly. it's all about association and the connecting and strengthening of neural pathways.

I used to get upset when I would forget an idea. Thing was, I could remember having the good idea but not the idea itself. Now, I simply get back into the same state, it will come back to the surface in its own time.

I don't think those thoughts get lost: I think they end up tuning your future thoughts.

Your reaction to those random thoughts is used as feedback to your subconscious and brain structure as to how it should organize and what kinds of thoughts it should generate in the future.

My model for notes is converging on Org Mode.

But... With a review process every day and every week, GTD-style. I see this need from my notes over the last couple of years from the job. Not usable for documentation (as I had planned) or helpful for other people.

I read about a guy at MIT(?) that could be mobile with a relevant setup years ago. It should be simpler now than it was for Mann et al. Just use an eyeglasses screen (Google, where are you?) and a chording keyboard strapped to your hand.

> But... With a review process every day and every week, GTD-style. I see this need from my notes over the last couple of years from the job. Not usable for documentation (as I had planned) or helpful for other people.

I've resorted to a big folder of paper for this very reason. I really want to do it all electronically, but my experience is for my own part I'm too likely to go off and start jotting notes in some random file somewhere, or test a new tool, and the notes end up sitting there, not getting referenced because I forget all the places I've made notes.

I've started looking at a computer screen as a window into a space that is otherwise largely hidden. The problem is there can be clutter everywhere you're not looking, and things get forgotten, or lost, unless they're regularly reviewed.

Paper solves that for me by being physically there, cluttering up my desk if I don't organise them, or in my notebook, making it easy do review and distill. Even when there's a lot of them, the physical bulk of it creates an implicit reminder to flick through it and see if its still relevant. Whatever is not still relevant goes in an archive for scanning.

Things that requires too much writing for me to want to do it on paper I still do electronically, but I'll make a paper note of it: e.g. it may be referenced from my list of projects.

I still hope to find an electronic setup that I can stick to consistently, but I think for that to work for me, I'd need to e.g. be able to dedicate a large enough workspace to be able to set aside a separate screen to keep my notes up and physically visible all the time (even if just as an icon on the desktop), or find another way of keeping them "physically there" in the same way.

I keep all notes in Org Mode, stored in a simple and shallow hierarchy in my Dropbox. (~/Dropbox/orgs/w for job notes. Etc.)

Since I have Emacs open anyway, the notes are just a few chars away.

(The best feature with Org Mode is that after I learned it, the note taking doesn't take much attention; it is transparent for cognitive load in a way that only a Pilot G-TEC-C4 + paper ever was for me. I only think about what I want to add, not how to do it. Ymmv.)

My main problem is to access those notes from phone/ipad/etc. I started https://github.com/berntb/attorg but got lost in the GUI/UX.

sounds like a typical case of "perfect is the enemy of good (enough)"

Looking at Twitter, a lot of those daily thoughts should be circling the drain.

But vacri, none of our minds create such worthless inanities! Historians will pore over our thoughts enchanted as if by beautiful snowflakes.


Spaced repetition is a wonderful tool. I came across it when I started learning languages and have over 5000 cards in my largest Anki deck for words and phrases.

A lot of thought is put into how to use a learners time most effectively. The algorithm designers seem to want to optimize their software so that a student only sees a card they are likely to be on the verge of forgetting. Because of this, uses often fail when they look at a flashcard.

Designing like this seems to ignore spaced repetitions most common failure mode: giving up. I think there is something especially demoralizing about getting questions wrong or being shown you are forgetting something all the time. I think it might even be solved by making the software less efficient but more 'friendly'. For example, giving a student an easier question that they are unlikely to have forgotten to soften the blow of a glut of difficult questions.

I remember earlier versions of Anki that implemented the algorithm so literally that it would tell you that you couldn't learn new cards until you had reviewed the 500 cards in your backlog, a seemingly insurmountable task that would just cause you to procrastinate and get a bigger backlog. They revised that to make it more motivating, fortunately, partially by capping the amount of backlog it would present to you on any given day.

I'm using WaniKani for Japanese now -- it's not nearly as portable as Anki but it's the first integrated solution I've seen for Japanese, teaching kanji by a Heisig-like method alongside vocabulary that uses the kanji.

I used to use an Anki deck for Heisig's Remember the Kanji, but I found it entirely unmotivating to spend time every day learning symbols out of context. Imagine an English class about learning the alphabet song, except the song just keeps going, and by the time you're learning the tenth verse of the alphabet you're still not spelling any words. That's Heisig. It's a clever system of mnemonics but it's not a complete solution.

Now WaniKani has the opposite problem with motivation -- it ramps up far too slowly. It doesn't let you study ahead. On your first day using the system, you can only study something like 20 cards, and you only get more when you wait out the spaced repetition and prove you've mastered those, even though they are extremely simple radicals and numbers that you probably know already if you are interested enough in learning Japanese to seek out WaniKani. This must be terrible for on-boarding, but I'm putting up with it because it's still the most effective Japanese study method I've seen.

Agreed. It doesn't matter how efficient the algorithm is if people get bored and stop using it. At Polly Lingual (https://pollylingu.al) we tried to tune our space-repitition flashcards for this. We aimed for small lessons that move more quickly, with optional follow up for difficult items. For casual learners this doesn't deter them, and for advanced students, they can still go back and see where they need improvement or just simply do the lessons again.

Fascinating, how come I never heard of your site? Is it fairly young? It looks interesting and a big selection of languages (I'm looking at Arabic).

But i just wanted to comment to mention that i like the pricing model - it seems small enough to be painless and still like a real price. Will definitely try it out.

Thanks, it's been around in various forms for a few years, recently moved from "Salaam Arabic" https://itunes.apple.com/us/app/learn-arabic-salaam/id646808... to a the unified Polly Lingual. It's been a lifestyle company and sort of priced that way. The goal has always been to not be annoying, but maybe that's been bad for business.

My main problem with spaced repetition programs like Anki is getting the motivation to stick with them long-term. This also dovetails in with my problems with procrastination. I've never seen a good solution. Most of the proposed solutions boil down to "just do it". It doesn't work for me.

> My main problem with spaced repetition programs like Anki is getting the motivation to stick with them long-term. This also dovetails in with my problems with procrastination. I've never seen a good solution. Most of the proposed solutions boil down to "just do it". It doesn't work for me.

If it's just one or at most three such'daily'things you want to do i had great success with simple "don't break the chain" kind of apps - every day you succeed in doing your task you check the box and it's marked on a calendar. There really is a drive not to ruin this beautiful chain, and I've found myself following through for months at a time - but then when you do fail once and the chain is broken the motivation really takes a slump as well.

Checking in with a friend or your partner can also help. Get some mild social accountability going and you might find yourself quite driven to follow through.

"If it's just one or at most three such'daily'things you want to do i had great success with simple "don't break the chain" kind of apps - every day you succeed in doing your task you check the box and it's marked on a calendar. There really is a drive not to ruin this beautiful chain, and I've found myself following through for months at a time - but then when you do fail once and the chain is broken the motivation really takes a slump as well."

This reminds me of the big physical yearly calendar I put on my apartment's door. It's right there in front of my face every time I go out the door. When I first got it, I crossed out all of the days that had passed for the year, and wrote in what was to come on important days of the year. Each day I would look to see what was on the calendar, and cross out yesterday's day. I had this nice long chain of crossed out boxes and written in appointments.

That worked for a while, but eventually, I stopped. Now it's been months and the calendar hangs there, and I don't touch it. I can't fail to see it all the time, every day I leave home, but I've just phased it out, I ignore it, and don't even notice it 99% of the time. Sometimes I do notice it or remember that it's there, and that I should really be crossing out those days, but I don't actually do anything.

That's the thing about procrastination. It's one thing to know what you should be doing, and quite another thing to do it. Tricks like "don't break the chain" can only get you so far.

I feel a bit the same. What's your goal of learning with Anki?

Part of my other comment here 'then what?' also links to someone else's comment about Anki - it wouldn't let you skip 500 words of backlog until you'd reviewed them. And the FAQ used to say something like: "Sorry, but you skipped some time and you are now at risk of forgetting those 500 words. You need to review them, because otherwise what's the point in any of this?".

I'm starting to wonder if people ever really 'learn' things, and if it's more a case of 'use it or lose it' happening every moment. When you're not using it, it starts to fade - whatever it is. Native language included, all skills included. And the consequence of that is that all learning has to include some notion of using it in future, not just once but regularly - frequently, even.

If your goal is in any way "because I should" or "because I want to be a better person" or "because I want to know more", I'm not surprised it gives you little motivation. If your goal instead includes "I want to do X with the language or skill, on a regular basis":

- I want to study Language because I travel to Langland every two weeks for work, and desperately want to be able to talk to people without a translator: motivating.

- I want to study Language because I saw a holiday to Langland on offer and maybe I'll go there next year, and I don't really have anything else to do: not motivating.

then the strength of motivation depends[1] on the strength of desire you have to do a thing, combined with the amount of doing that thing you want to have in your future life. Most things we choose to learn about, we have no good plan to weave them into regular life, and so not a long term strong desire to learn about them, and so not that much motivation to stick past the novelty stage.

[1] speculative.

I used this to memorize/train problem solving (applied to compsci problems).

I solve a problem at SPOJ [1], Codeforces [2], or project euler [3] or other similar sites. Put it as a card in Anki and revisit it after it appears again.

Solving the problem for the first time requires some research into algorithms and methods (took as much as 4 hours when I was just starting), and solving it again is remembering the tools.

Nice thing is that the evaluator tells me if I'm right or wrong, or if it takes too much time I've obviously forgotten how it's done.

I unfortunately stopped doing it as most problems take more than a minute to solve (each needs about 10-150 lines of code [depends on problem complexity]) and I eventually had to do 10-20 per day. At a particular moment in my life I didn't have time. I'm thinking of starting it again though as I've noticed my coding was much more fluent.

[1]: http://www.spoj.com/

[2]: http://codeforces.com/

[3]: https://projecteuler.net/

I do something a little different but only because it's simpler and doesn't really require any specific programs to use. I dump all my journals, diary entries, notes, thoughts, trivia and miscellaneous things I would like to remember into a giant text file. Then a few times a week or whenever I feel like it I ask the web to give me a random # (numbergenerator.org) and that's the line I go to in the text file and read from there. Sometimes I also just browse around the file at random and read away as well. I don't have any statistics on my retention rate of course but generally I think it helps me recall things better and amongst my family and friends people occasionally comment on my good memory and it could be this method is a part of the reason for that. I think in principle the concept is very similar to the spaced repetition method, perhaps just a bit more unstructured.

I wrote a program I fancifully called 'Human Unit Tests' to aid me in my studies (learning a diverse set of constants for biophysics). I can very much attest to the effectiveness of spaced repetition.

But, /boy/, do you need to stay on the ball. You can't really afford a cavalier, let's-see attitude with this (given any non-trivial amount of items-to-be-memorized). The review process needs to be as much part of a daily routine as workouts ... Yeah.

On the other hand, there's one reward that doesn't usually get mentioned (as in the fine article re-submitted here[0]): the strengthening of corollary knowledge (or coordinate terms, for the linguistically inclined).

[0] Previous submission: https://news.ycombinator.com/item?id=5809762

Can you expand a bit on the coordinate terms issue? My impression would be that comes rather from the encoding rather than recall?

Sure, I'll try to elaborate.

Suppose you're reading a biography of Huygens. You may find yourself inspired to memorize a few of the basic facts therein. Dutifully, you feed his life's dates, his major acquaintances and maybe a few places of importance into the SR system of your choice. You are committed and keep repeating those facts in ever-increasing intervals.

After a few years a random conversation touches upon the very subject. To your delight you discover that you are able to hold forth on Huygens, the man and his time.

To your surprise (and this is my contention [and experience]), you also find yourself able to speak with some level of accuracy about tangential matter -- eg. the theories he worked on -- without ever having either added related facts to the database or dealt with the subject matter in the intervening years.

In other words: recall of a whole web of interconnected pieces of knowledge may be strengthened considerably by spaced repetition of just a few of the central facts.

In my experience there's no specific 'encoding' procedure necessary. I never put any thought into carefully selecting facts for the spaced repetition treatment, yet the effect usually manifested itself. So, yes, I would say it's a 'recall' phenomenon inasmuch as the brain does all the heavy lifting.

Fascinating thanks, this is a new term for me but strikes a chord.

It also fits nicely with the limited understanding we have of the recall of information in our brains - it all comes down to context and activating the right network (or paths) which can only be reached by activating related/overlapping networks. So once you activate memory on a specific issue you can more easily activate related information (or even do so without intention as you describe). Having more easily reachable 'access points' (strongly encoded and thus well connected information) makes it then easier to access related information.

A corollary is that in order to remember information it's important to connect it to previous well established memories (eg "how does this new concept fit my own experience").

Thanks for this.

One thing I felt I discovered over time is that if you wish to retain all knowledge you learned in a course, it's basically impossible to do so at a normal course's pace. The courses move too fast. You basically have to cram and forget.

I based this off of a Coursera course - I made anki cards for all the new information and made sure to review all the new cards on the day I learned them. The review time rapidly turned into hundreds of cards per day and it just didn't scale.

For that one Coursera course, I ended up being stubborn and limiting my pace to what I could Anki. I ended up taking the Coursera course three times - I passed 100% the final time, but still hadn't finished putting the last two weeks of content into Anki.

I've dabbled with this and I find the hardest part is deciding what's important enough to memorize. Most rarely needed facts can be looked up as needed. Commonly used information will get enough practice naturally without spaced repetition. You have to decide what's important even though you don't practice it naturally.

The obvious exception is learning a language. And if you're in school then this is decided for you.

>Most rarely needed facts can be looked up as needed

Unless you don't remember they're there to look up.

Whenever I make cards about a programming language feature, I suddenly find myself in situations where that feature is perfect. Either I'm the luckiest programmer on earth, or else the brain has some clever tricks up its sleeve.

There is an amazing slightly older longform article on the creator of supermemo (the original spaced repetition software, although the concept of course comes from a German psychology pioneer of a few centuries ago (Helmholtz i think?)). The guy throws his whole life in the software, eg every news article, email, ... Is just processed with cloze scripts (= "fill the [...]") And his aim is to remember everything. The supermemo site also has a lot of fascinating articles written by the guy, really worth looking at if the topic interests you.

https://www.wired.com/2008/04/ff-wozniak/ pretty sure that's the article you're referencing

exactly, thank you!

I used this at work; while waiting for builds to complete I would use Anki [0] to learn all the names and locations of the countries. It worked pretty well, but it isn't a silver bullet. I retained maybe 50% of the information.

[0] http://ankisrs.net/

>I retained

You're doing it wrong. You're supposed to continue using the deck, indefinitely or until you're okay with losing it all.

When a deck's life is measured in years, most cards' appearance frequency also become measured in years.

I found it really motivating to see the message that a card will appear again in 4 or 7 years! But i never got beyond a year of actually using a deck...

Helps if you have a daily commute on public transportation (like in NYC or London or something). Perfect way to make that time productive :)

What deck(s) do you use? I've been looking for a deck that ideally has one side where there is an image of the country, and I give it the name, and another where there is the name of the country and I click on it in an image, but it's hard to find such a deck.

Some people swear that the process of making the deck helps significantly with their ability to retain the material.

Personally I find it a waste of time and all time spent editing cards is a waste to me. It doesn't help me. Learning traditionally (taking notes, answering practice questions, etc) from books and reviewing with Anki is the most effective for me.

Ymmv but when building a deck creating mnemonics for each item was really important for me. Made it a lot slower to enter data but much more fun and much more effective in later use.

I found mnemonics to really only be useful when you're in the beginner stage. Once I got up to intermediate real world usage was much more helpful and easier to remember since I didn't have to remember a completely unrelated thing along with what I wanted to learn.

The highest rated geography decks was what I used [0]. But yeah finding good quality decks is the hard part, though the geography ones generally do have pictures of the country and you guess the name.

[0] https://ankiweb.net/shared/decks/geography

TinyCards from DuoLingo has that feature.

.. and is an iOS only SRS card app.


Anki has various shared decks to download that have things like name, place on map, flag, etc

I believe SRS is overused. The spacing effect is, of course, real and quite important. However, it lends itself to decontextualizing what you're learning.

For example, foreign language vocabulary is best learned in context (ie., in sentences, within conversations). That's the only way to know how it's actually used. Otherwise, you're just learning a mapping between first and second languages.

There are other benefits to learning in context, as well. It's more fun, sustains long-term interest better, and these emotional benefits enhance memory as well.

Most interestingly, there are efficiencies to learning in context. For example, one can learn two new words in one sentence.

For all these reasons, I believe it's quite unfortunate that the first language learning tool everyone grabs is some kind of SRS flashcard system. And similar arguments apply in other domains.

Spaced repetition and learning in context are not mutually exclusive. [0]

Secondly, there are many many advantages to learning vocabulary out of context. I am very interested in foreign languages, and have self-studied to elementary proficiency in Italian, German, and Russian. Memrise is my go to app for picking up vocabulary out of context, because it is the fastest path to a bigger vocabulary. For a lot of vocabulary, at least for the languages I've come across so far, there is a direct one to one mapping for the essential things.. E.g. you can learn the word for "book", "fire", "box", "sky", etc. out of context. Certainly, learning in context has advantageous for improving grammar and in some situations verb usage - therefore a more balanced approach. An easy example of the advantage of in-context language learning is that in German and Italian, knowing somebody uses different verbs than knowing a thing. If you were to study a verb like "to know" in Memrise, you rely on the flashcard's definition to be thorough enough to clarify their usages.

[0]: "A Trainable Spaced Repetition Model for Language Learning", writes about how spaced repetition is used by Duolingo. https://s3.amazonaws.com/duolingo-papers/publications/settle...

Thanks for the link to that paper -- I hadn't seen that. I agree that SR and learning in context are not mutually exclusive--but they're challenging to integrate. Duolingo is not a good example -- they teach through random sentences, out of context.

Your examples - concrete nouns - are where you miss the least by simply mapping onto English. But I would suggest that this is the very easiest part of learning a second language.

I think many learners underestimate how much of language is not derivable by plugging vocabulary into grammar rules. There's so much that is just "that's how we say it". And you need to pick that up in context.

SRS is most effective when used to retain what you learn, not as a way to learn new material for the first time.

This is the subject I tackled in grad school. It was super interesting, because many people still believe that memorizing stuff is best if you cram for an extended period of time. Showing them that just a few minutes of review at properly spaced intervals leads to better memorization floored most of the participants.

My particular research was the quantify the effects of the variance in follow-up exposure times. I did this research 10 years ago before smartphones were available. The participants had to physically be in a classroom in front of a computer to get the treatment. So in our experiment's case, the effects of variance were non-negligible and needed to be investigated. If smartphones had been around then, we could have tested so many more hypotheses.

The existence of spaced repetition also shows how dysfunctional educational systems (both public and private) are at making evidence-based improvements. If a method with such a clear overwhelming advantage at the stated goal isn't incorporated by schools, what hope is there that they'll make improvements where the evidence is hazier?

One of the most cited papers in the area is Frank Dempster's paper about the failure of educational systems to adopt spaced repetition: http://andrewvs.blogs.com/usu/files/the_spacing_effect.pdf Although the paper was was written almost 30 years, little has changed.

It is taken up, just not everywhere and not necessarily at scale. I've personally seen it used eg at Finnish schools - but there the teacher chooses the tool, not the government, so i don't know how common it is.

With more and more devices in class you can expect directed spaced repetition to hit hard. But actually i don't know a single person who didn't at some point get the instruction/advice to use flashcards for foreign language vocabulary. That being said it's not used much in other subjects - because no serious educationalist thinks education is about hammering facts into the brain. It's much more about concepts and critical thinking, at least in any semi-modern classroom .

Cramming might actually be the most time-effective strategy for a short recall period (one off exam). But naturally not if the aim is long term recall.

It is. We were specifically dealing with ESL/second-language learning though. Long term retention is necessary for fluency. If you want to pass a driver's exam or material you don't really care about beyond passing a test, then cram. Although in most classes, you will probably want to retain knowledge beyond a month in order to do well on a final.

can you compare this with memory palace and other, alternative techniques, such as using the information to be learned actively and immediately?

I've tried using Anki but found it too confusing. Most features are overkill for what I needed, plus I always forgot to revise.

Earlier this year I build a side project to tackle this problem. I've added email notifications, made interface really simple. Only left what it is - spaced repetition app - Endue (https://endue.me/). After a month of using it, I actually implemented an add-on - Telegram bot, EndueBot (https://telegram.me/EndueBot).

Telegram Bot is super useful as you can revise without leaving Telegram and most of my cards don't require audio or images, it's text mainly. It only takes few minutes a day.

I've been using it for the last 6-7 months, and it helps me with dev cards. For example, if I'm doing a project and need to google some error or explanation on config option or whatever, I add a card after I find an answer. So next time you have this problem instead of googling you just recall it from your head.

There is Messenger version in progress, I should finish it later this month.

Would love some feedback on it.

Looks interesting, might give it a try if I can put together the spare time to start learning a language.

I am a big believer in the effectiveness of SRS for getting a basic foothold in a new language or subject, but I stopped adding cards to my Anki decks after awhile because the effort needed to create high quality cards is too high. Studying is a breeze with SRS, but inputting and formatting study material into a computer takes too much time, especially if you want to include images or other context. SRS works best with small atomic units of knowledge that stand on their own without context.

I know of no better way of bootstrapping those first 2000 words of vocabulary in a language, but beyond that I think you're much better off using extensive reading of easy texts to build on your language foundation.

"I know of no better way of bootstrapping those first 2000 words of vocabulary in a language"

For me, a much, much better way has been something called Total Physical Reponse (TPR)[1]

It basically involves a speaker of the target giving the learner a verbal command to perform an action, demonstrating the commanded action, and then prompting the learner once again with the same command.

For example, if a teacher was teaching English, he/she could say:

"sit down"

then demonstrate what they mean by sitting down themselves, then say "sit down" to the student. The student would then mimick the action of the teacher, by sitting down.

Then the teacher could say, "pick up the fork", pick up a fork themselves, and then say to the student, "pick up the fork", and the student would pick the fork up, etc.

This is super effective for learning vocabulary and grammar at the same time, in context. For me, that's a much easier, faster, and more natural way to learn than using any flashcards or spaced-repetition program.

The downside, of course, is that this is just for listening and understanding. It eventually will have to be augmented with other techniques for speaking and learning to read and write. But for just learning to hear and understand the basics of a language quickly, that is the most effective technique I know.

[1] - https://en.wikipedia.org/wiki/Total_physical_response

The first time this hit HN I did an experiment of learning the US state capitals with spaced repetition. Within a couple of weeks, I knew them by heart. So I stopped, and now can barely remember any. Memory is a funny thing! The very infrequent revisiting of stored knowledge beyond the initial learning does appear to be critical.

If this image is correct then after a couple weeks the forgetting curve is still fairly steep: https://www.gwern.net/images/spacedrepetition/forgetting-cur...

Med student + developer here. As one that has to memorize insane quantities of information in a limited amount of time, I'll offer some thoughts on my experiences with various memorization strategies mentioned (or not mentioned) in the article.

1. Understanding the "big picture." The idea is, if you understand the mechanism underlying X, you can reason out Y. This works great for physiology-type where things work in predictable ways. Do I need to memorize pulmonary function test results? No. Because I can reason my way through the consequences of different pulmonary diseases. However, this simply does not work for things like anatomy or pharmacology where there are thousands of names of things to memorize (many of which don't provide any clue in the name itself). Yesterday I heard from a friend that worked in a related industry that pharm companies typically pick from a list of randomly generated syllables to choose a name for the generic of their drug. There is incentive to pick an unsavory, unmemorable amalgam of syllables to discourage people from buying the name brand once their patent runs out.

2. Acronyms are sometimes helpful but mostly overused. There are acronyms for literally everything in medicine, to the point where each new acronym runs the risk of being noise. The most helpful ones usually have to do with ascribing the order to things. "Some lovers try positions that they can't handle" for wrist bones or "Oh, Oh, Oh, to, touch and feel very good velvet AH" for the order of cranial nerves. A great example of a worthless acronym is the TORCH infections, which stands for "Toxoplasmosis, Other, Rubella, CMV, and HSV." That "Other" stands for the following: Coxsackievirus, VSV, Chlamydia, HIV, HTLV, Syphilis, Zika. Super helpful... :P

3. Method of loci. My first block of medical school, some two weeks before exams, I bought that oft-touted "Moonwalking with Einstein" book. It was a bit of an act of desperation. I had previously heard the author speak at Railsconf and figured I'd give the whole memory palace a try. I read the book (well, the pertinent parts) and spent a couple days figuring out how to encode chemical names into a person-action-object system. Then I translated the steps of carbohydrate metabolism into a walk around my college campus. Results? It worked! However, it took WAY too long. I could've done the same thing in less time by simply drawing the steps out on a sheet of paper over and over again.

4. The testing effect. Way back in high school, my AP psychology professor gave us reading quizzes before every class based on the studies that showed testing improved recall more than homework assignments. To this day, I remember more from that class than any other class I've taken, even those in college. I wish more professors used this strategy, but it takes more time and prep on their end. My particular medical program only tests at the end of each block in the form giant (8-hour) cumulative exams. On one hand, it's nice to not have to worry about balancing midterms or tests from multiple classes the middle of our eight week blocks. On the other hand, there is no way to benefit from the testing effect.

5. Spaced repetition. I agree with much of what the author has to say about spaced repetition. Yes, it works. But there are two large issues with it. The first one is having the discipline to regularly rehearse the information. Pounding flashcards everyday gets monotonous (and sometimes discouraging). The author argues that spaced repetition prevents you from having to review too much at a time. But that depends on you keeping a pretty strict schedule. Miss a day, and you have double the cards to look at. Miss a couple of days and the amount of cards you have to drill to catch up becomes psychologically prohibitive.

The second issue is that flash card creation is the absolute rate limiting step. Every flash card program that I've looked into (Anki, Quizlet, among others) has a terrible UI. Frustrated by this, I actually spent ~100 or so hours of my first med school block developing my own SRS program. Even with some improvements made to the interface, flash card entry was still too slow, to the point that I stopped using after that first block. A year later, I dusted off my program and I added a feature that allows me to import questions from Workflowy notes taken during lectures. This proved invaluable in studying for my most recent exams (currently using it now). I can keep question-like notes in a readable format on Workflowy, then do a massive import to build my cards.

Another minor issue is that there is no way to value one fact over another. There should be a way to mark something as high-yield so that it appears more often. Sometimes I am faced with a decision of, "Well, I am running out of time... I can drill these cards, or I can open my notes and try to pick out the highest yeild information."

I have a lot of thoughts about improving upon these prohibitive issues with SRS. I'm hoping to roll out a more comprehensive solution next year when I have more time to code.

FWIW here's a screenshot of my current software: http://i.imgur.com/sVTlMvP.png

Very interesting thoughts and my experience is very much the same!

One question: do you still remember the chemical path today (vs other similar items)? My experience with loci is very positive but i agree it's time intensive. From what I've read practice makes it quicker, but it is actually this time spent and the deeper processing and linking with other information that occurs when you use loci or most other mnemonic methods that has the memory-strengthening effect.

I don't remember much of it. I think it's because the PAO system was too complex. I can remember some of the images, but I can't remember what they stood for. I have been able to recall some abstract things (like where other pathways like glycogen storage intersect with glycolysis). What have you used loci for? I was thinking I might have better luck using loci for pharm stuff. But Sketchy Pharm exists, so why spend all that time when they'll create scenes for me?

Shopping lists and to do's mostly, but the to do's got unattainable as they change frequently

>The second issue is that flash card creation is the absolute rate limiting step.

Supermemo has a very simple Q&A text format that you can import. I typically write everything I want in Vim (in a format that makes sense to me) and then run a script over it to write it into "Supermemo format" that you can just import from the GUI. There is a more advanced XML format but the simple Q&A text format has always worked fine for me.

I really wish Supermemo had a Web interface though. I'd love to use it from my mobile while I'm having lunch.

When growing up, rote learning was one of my main gripes with school. Spaced repetition is rote learning with controlled (instead of brute) force.

With time I started to respect memorization for certain kinds of information, to the point of using it on my own vocabulary learning app[1], which is one of the few cases where memorization seems justifiable, but one still has to put the information to use, otherwise the info will be lost.

[1]: https://itunes.apple.com/app/id1126547471

Spaced repetition flashcard apps are the "Smoke on the Water" of developer side-projects. Here's my attempt: http://cardflashapp.com

Someone should make a repo of all of them and then the best techniques could be compared / contrasted for future attempters.

> the "Smoke on the Water" of developer side-projects

What do you mean by this?

That is popular among guitar players to start with "Smoke on the Water" as an easy song to learn basic principles.

I've been using Anki to learn (English) words for a while now (I got tired of looking up words I come across in books and then not 'learning' them from that). I've noticed a few things:

* A short definition is harder to remember than a long, involved one. It feels like the short one doesn't give enough mental 'hooks' for connecting to other concepts.

* A word that sounds onomatopoeia-c is usually easy to remember -- a 'cosh' is a bludgeon, and I have no trouble remembering that because it sounds kinda like one - one-syllable, harsh, blunt

* Words that don't seem to have definitions related to their spelling are hard to remember -- I had the hardest time remembering the meaning of the word "prevaricate" - 'to speak evasively, skirting the truth', because 'pre' makes no sense in it and 'vari' is not a root I've heard before, and neither has anything to do with lying

* Learning the etymologies of complicated words helps a lot, especially if it fixes the previous problem -- after learning the root 'aestus' means fire/boiling, but came to refer to the tide as the 'boiling of the ocean', I can't forget that estuaries are tidal zones of water

* Learning a story behind a word makes it hard to forget -- for instance I remember the word 'boffin' (Britishism for 'nerd') because I remember how it was used positively to refer to the war-effort researchers in Britain in WWII

* Using the word in a reasonable sounding sentence (not an idiomatic usage, 'the one way the word is ever used') helps a lot -- for instance, find a way to use the word "mores" that isn't in the phrase "social mores" and it's going to stick a lot better

* Words that come packaged with connotations are easier to remember -- for instance, the word 'lucre' means money or profit, but knowing that it usually means 'dirty money', specifically, as in money obtained via crime / sordid means, makes it way easier to associate it with a specific concept

* Talking to someone else about a word I learned makes it many times easier to remember -- for instance I gushed to a friend about the word 'psithurism', which is the 'sound of rustling leaves', and can't seem to forget it now

* Using a word in conversation / my own writing makes it far easier to remember than any amount of studying it via flash cards.

Basically it feels like the actual way to really learn a word is to associate a big mental structure around it in my mind - to understand subtleties in the definition; to associate it with its parts and roots and history; to have a social memory of sharing the word with someone else; to have to do the mental calculations to use it in my own speech or writing and to decide that it's the best word for what I wanted to say; etc.

This is consist with my experiences learning math and physics - I can learn a subject if I see it as a big picture, a narrative I can re-develop on the fly, or as a package of tools that are motivated by a particular kind of problem. I can remember an equation if I associate the parts of it with complicated mental frameworks or detailed arguments. But reading it off the page a bunch of times gets me almost nothing.

Cramming for a test usually means stuffing the raw text of the material into your brain, because you don't have time to build the whole concept lattice around it. I think that's why it doesn't work for retention.

I'm always happy to see new stuff from gwern. If you like his posts as much as I do, you should contribute to his Patreon: http://patreon.com/gwern

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