Ask HN: How do you memorize things you read?

115 points by boppo1 ↗ HN
— especially if they're technical you don't use them regularly at work?

For example, I don't work in finance (got the degree but my grades were abysmal), but I have a great deal of interest in the field, so I bought the CFA and CAIA curriculum. As I read it I grasp it and answer provided problems with no difficulty. But when I return to the material a few months later, I may as well re-read the whole thing for the amount of detail I've retained.

98 comments

[ 3.2 ms ] story [ 219 ms ] thread
(comment deleted)
I train myself to look things up in the manual quickly. For instance I use four programming languages regularly and sometimes have to use another. Sometimes I forget how I'm supposed to, say, look up the length of an Vector/List/array but it's OK because I can look it up in a few seconds.

When I do cardio at the gym I take my tablet and read technical documentation, frequently over and over again. The goal of that is not so much to memorize facts, but to become familiar with the landscape and to become very quick to look details up when I have to look them up.

I've yet to see a question in a data science interview that can't be answered with: (1) Look it up in the hashtable or (2) Look it up in the literature.

Yeah but this is backwards. We're remembering how to remember something instead of just actually remembering the thing.
Are you sure you want to memorize it? It seems better to understand it well.

If you do want to memorize it, then some form of flashcards with spaced repetition.

If you want to understand it, keep reading and dig deeper into the topics that interest you. Also for technical stuff work through some examples.

It's very hard to actually comprehend any technically challenging concept without being constantly exposed to it. The more challenging it is, the more time you're going to need so it become something like second nature to you.

Take up as many relevant projects as needed to master the underlying theoretic aspects of whatever you're interested in, whether that's in math or programming. Don't get frustrated if you get stuck somewhere. You can always take a break and jump back to take a better look to any helpful readings to better understand some concept you may have missed earlier. Just make sure you're already pretty familiar with the very fundamentals first (e.g. don't expect good progress on a machine learning course without being already familiar with linear algebra - probability fundamentals.)

As I was writing my answer you posted yours, and I found It very very helpful.
Modafinil.

Aside from that, for me, it's more important to memorize where on the page I saw something, and where the book or page was located. Things like web pages may get saved because of their ephemeral quality. It's unrealistic to expect to retain everything, but knowing where to find non-critical knowledge is easier.

This is great when you can access the book. Great for info I don't want in my head. There is a cost to this however. I feel like the brain has a few slots for temporary into. And those slots are also needed for problem solving(holding temporary mental models and theory's to test). It works, and is faster for the learning process, I just feel it's slower an the application side.
I usually need to find "usable" ways of introducing the subject I'm studying in my day-to-day life.

As for finance, eg. I'd try to read Economy mags more often, and try to explain to myself just "what the actual hell" they're saying.

This method can be very challenging when you're starting on something, but by the time (two weeks) you get to know "just enough" you'll be flying through the news, and then you can get to scientific papers on the subjects that you have interest on.

For programming, my way for it is exactly the same, but without newspapers and articles: just try to approach everything in a specific mindset.

Maybe your mind cannot be applied this way. I recognize you are being proactive by asking here, but the people you are competing against do not have this problem.
Is there some kind of a study that people can use to benchmark their finance textbook remembering ability? Because after factoring in a small amount of exaggeration in "might as well not have read it," forgetting a substantial fraction of a million facts, all of which the reader found boring, after two months with no reinforcement, sounds kind of plausible.
I have considered this. However it seems nonproductive because it just encourages me to give up. Better to assume it's possible and try.
Okay, well the flip side is that the CFA pass rate is 25% so most of the people up to that point do have issue with both retention and applying what they learn

Curious to read other people’s strategies here

Recently I started learning how to speed read with the help of an app that quickly flashes a few words at a time.

This type of reading has forced me to re-think how I memorize or even check that I've understood something.

My current process is to stop at the end of a chapter and re-cap what I've learnt and id necessary check by reading in normal mode.

Train the recall neural pathways for the scenario you envision needing the info. Active recall at intervals.

This is the only thing that really works for me.

What I have not automated is creating active recall scripts as I go....it tends to happen organically as I run through scenarios in practice sessions, eg if I want to share info verbally, then what parts do I forget, when I practice? Drill on that section, and test it out ant intervals until it's locked in. Then test at much wider intervals to maintain. (Super memo and Anki have complex algorithms, but I like to ballpark it to reduce complexity)

I tend not to memorize in that way. Reading gives me a basic familiarity with a topic and, more importantly, helps me understand where to find information when I need it. If I use a piece of technical information regularly, I'll just keep looking it up until exposure fixes it in my memory.

However, if it's something I do need to remember and can't look up, spaced repetition tools like Anki are helpful.

You should probably ask yourself whether it's worth the effort to memorize something that you only need every few months. So long as you have an understanding of the general topic, it's not hard to find technical specifics when you need them. Being able to reel off technical details from memory is a good party trick, but I've rarely found it to be useful in daily life.

Yup, same here.

I absolutely don't memorise things on purpose. I just use them, and try to remember where to look them up again.

Memorisation is a fancy tricky that doesn't gain you that much over knowing where to find the answer.

(comment deleted)
No need to memorize everything in this day and age. Memorization is super useful, but if it's not part of your day to day, looking it up when you need it is a more efficient use of time.

To that end, I'd instead recommend taking notes in a system that allows for full text search.

However, if you still want to memorize it, repetition will be your friend. Repeatedly reading the material will gain you familiarization with the concepts, and eventually the memorization. Spaced repetition (Anki et.al.) is the fastest, but simply reading regularly works too.

I have a "trick". My native language is not english. 99.9% of what i read is in english. Comprehension for me is also slow when just reading(at least once).

Memory, as i found out, works if you connect new knowledge to existing knowledge. Usually, when i read and understand something to the point of the "aha" moment is not sufficient. Just understanding something in that moment is not enough for storing in long term memory.

Here is the trick part: when i read and want to comprehend and memorise, i do the following:

  1. Read a paragraph/piece in english
  2. Get to the aha moment by re-reading until it makes a narrative in your head
  3. Once i have the narrative, i translate it to my native language and write it down(pen and paper > keyboard)
It's not fast, but there are many processes involved here that get to long term memory:

  1. first reading - getting the context of the material
  2. getting to the aha moment - means it's connected to something, and the new knowledge somehow filled the gap
  3. when i get to the narrative part, i can make a story and tell it to myself, in my own head "this is because of that etc"
  4. #3 is not sufficient, so i take a step further and process it by translating it - further cementing the "connecting to old knowledge" process
  5. #4 is also not sufficient, as the translation is not always clear when trying to explain it to someone else. here is where the longhand writing comes into play - writing is expressing your thoughts clearly. when this is done, i already went through a lot of phases of processing, that i can say i fully understand something without any rote memorisation: i can just use whatever old knowledge i have to recreate new knowledge and to assemble it based on the listener's (perceived) capabilities of understanding.
I know this works, because i have all the processes written by hand, in a notebook. Everything i explained here is a narrative that i constructed on the spot to explain my process. Keep in mind that this whole process of comprehension, memory, new knowledge and handwriting was new to me ±1 year ago.

Not sure this helps you directly but indirectly it should be able to guide you into figuring out what works best in your context.

  #edit: i still don't know how to format text in a comment

  #edit 2: figured it out
To retain facts and terminology, do you simply translate to native language to retain it?
This is a good question. Facts still fall under the context. Think about 1989 as a date - you know a lot of facts around that date. Think now of 1986(Chernobyl) - you also know a some facts about this date. 1504 - not so much(i assume). 2006?

The point is that you can also retain facts if you can connect them to something else. The more link you have the easier it is to retain.

For terminology - this is tricky. Whenever i run into some new terms that i cannot explain into something else, but need to learn the term, i try to use it in written text as much as i can. But whenever i try to refer to it, i refer to it from the explanation - hoping it(brain) will bring me the term. If i use it a lot, it will get cemented.

Take a look at incremental reading -- the big app in this space is supermemo, but there are plugins for Anki (another app) that might make it work. Incremental reading is great because aside from tackling remembering things you've read, it also helps schedule what you're reading, so each day you have a 'session' and then what you read is decided by the program.
Don't think Anki has a fully baked implementation of incremental reading.

Polar [0] is an interesting implementation of a similar concept: read and annotate and turn your highlights into Anki flashcards automatically.

[0]: https://getpolarized.io

Readwise (paid service) does this for me.

Lots of integrations (kindle highlights, hypothes.is highlights on the web etc.), and they present your old highlights to you in the form of spaced repetition, either via email or via the smartphone app.

If I really need to remember something I use spaced repetition, Anki being one of the more well known tools for that. Spaced repetition is particularly useful if you need to retain information indefinitely; you can do so as long as you stick with the process. The downsides to it are the upfront cost of ingesting whatever information you're looking to memorize into what is usually a flash card like system.
I don't memorize; I try to learn a skill by applying it in practice on some project, no matter if real or a "toy" one; most of the skill will stick, but the memorization of all the facts facts is rarely relevant - any facts which were not required to apply that skill don't really need to be memorized and can be looked up easily when/if needed.
One of the most successful methods I've found for me is to rewrite the material. I haven't seen anyone suggest this yet.

By rewriting the material, you are recoding it. You are forcing yourself to think about the information differently: what you are doing is explaining it to yourself. This to me is far more effective than something like highlighting, because in rewriting that material, you are also clarifying your understanding.

This is a great suggestion. It worked quite effectively during my high school while preparing for college entrance. I used to rewrite a lot. Rewriting in combination of revisiting material with some frequency helped.
I do this all the time. I write things down and never come back to read them because I seem to remember the important parts.
Ben Franklin did something similar, he copied books by hand I think
I think he rather took short notes, then attempted to rewrite the material based on the notes, comparing to the original afterwards. This would seem to me to be much more effective than copying directly out of the book when it comes to recall.
Very much this for me. It's the reason I wrote notes instead of typed them in college. The intention was never to use the note for reference, but the act of writing them enforces your memory.
I'd think that for code, typing them is more effective than writing?
I've never written down code for school notes. At least nothing more than a few lines. There is no point.
Yes, this is my approach too - I'll take extensive notes and/or rewrite the material. The way I do thinking and learning is essentially to write things down on paper - it doesn't work for everyone, but it does for me.

I also happen to use fountain pens for it, but I think that part is optional. :)

I can vouch for this. I just gave an SDE a bad performance review because he could never recall any code he worked on or how it worked…
I wasn't sure so had to ask. Where you making a joke? Did the sde have actual measurable perf problems?
The lack of recall led to severe perf problems. He had a hard time fixing bugs quickly on code he had written. A significant change from the past, I presume he has started smoking a lot of weed.
Yeah that is sad. Software engineering for the most part has become "standardized" and it is unfortunate this individual either missed the boat on the upskilling or just ran out of steam/luck from what may have been (possibly successful) cowboy coding days!
This sounds more like they’re burned out. Subordinates going from great to terrible usually means the business is failing them, not they suddenly “started smoking a lot of weed”. I would take a hard look at your job as a manager.
No, he smells like pot.
Perhaps you should consider your part in that. Or if you could have had a more positive impact on his life apart from kicking him when he was down.
Unfortunately we do stack ranking so it wouldn’t be fair to anyone else on the team. He also refuses to say anything in 1:1s with regular nudging.
He might have had COVID as well.
This can also be caused by persistent sleep problems, which a lot of people have had during the pandemic.
Is there scientific evidence linking marijuana to memory loss ?
Yes, although as always with drugs I am sure plenty of people will say that all those studies are flawed or biased.

You can see it change rat brains:

> As people age, they lose neurons in the hippocampus, which decreases their ability to learn new information. Chronic THC exposure may hasten age-related loss of hippocampal neurons. In one study, rats exposed to THC every day for 8 months (approximately 30% of their lifespan) showed a level of nerve cell loss at 11 to 12 months of age that equaled that of unexposed animals twice their age.

It has measurable effects on human performance:

> Among nearly 4,000 young adults in the Coronary Artery Risk Development in Young Adults study tracked over a 25-year period until mid-adulthood, cumulative lifetime exposure to marijuana was associated with lower scores on a test of verbal memory but did not affect other cognitive abilities such as processing speed or executive function. The effect was sizable and significant even after eliminating those involved with current use and after adjusting for confounding factors such as demographic factors, other drug and alcohol use, and other psychiatric conditions such as depression.42

https://www.drugabuse.gov/publications/research-reports/mari...

Oh gosh, I am pretty bad at this. The only thing that seems to help is being on keto, in which case my recall is amazing.

Never got a poor performance review for it, but that's a real issue I have.

(comment deleted)
+1

This is what I do for most of my university courses. It's super not efficient time-wise, but my grades have been generally good thanks to it.

Plus if I have to write an essay about anything related I can just ctrl+f the material I've written so far and get my facts straight very quickly.

Key to memory is to use it often, if it’s not used it’ll be replaced by new stuff.

Speaking from experience. when I stopped using a third language and learned a fourth, the third is now almost gone, replaced by the new language that took its place while first and second languages that are in constant use were not affected.

People are giving app suggestions (supermemo, Anki), which is great, but the underlying concept is "spaced repetition" [0]. Tools like Anki (and supermemo, I presume) give you flash cards at an interval based on the spaced repetition idea.

Our memory is structured so that we forget things at an exponential rate (exp(-a t)). The better we remember things, the smaller the coefficient (a). You can look around for pictures of "forgetting curves" [1].

The idea is that for new knowledge, the coefficient (a) is large, so you need to refresh your memory sooner. As you retain the knowledge (the coefficient gets smaller), you can wait longer before you need to refresh your memory. The exact coefficient is going to be dependent on the person, the type of knowledge (for the person), the context in which it's learned etc. etc.

A simple way to do this without flash card applications is to try to recall the information you've learned periodically. Can you recall it after a day? A week? A month? If you can't, go back to the information and relearn it, and try again, making sure to do active recall to refresh your memory.

Another trick is to try to explain the knowledge you've learned to someone else, or pretend you're trying to explain it to someone else, if you don't want to subject your friends to it. Often times a "why" will come up or some piece of it will seem arbitrary that you won't be able to explain, which is a good signal that there's a concept you've missed and need to review to understand it more deeply.

[0] https://en.wikipedia.org/wiki/Spaced_repetition

[1] https://elitemedicalprep.com/what-is-spaced-repetition-and-w...

One of the best articles on SRS out there:

https://www.gwern.net/Spaced-repetition

That is an incredibly beautiful website. Thank you for sharing!
Gwern is like the harvester combine of document review, when he has passed there is nothing left, just some dry stalks. I've learned so much just by looking at how he organizes his work.
I’ve been using SuperMemo for almost 16 years, every single day. I can’t think of a better way to guarantee that I will retain stuff outside of using some kind of algorithm to tell me when to refresh desired information. I think of it more as “computer augmented memory” rather than “I’m letting the computer think for me.” I use it in the morning to review my cards but other than that I have little interaction with it (until I add more cards, which happens during periods of down time)
Organization & Repetition.

Any material which is worth knowing, but that will not become part of a routine (i.e coding patterns or structures that would be used everyday) which would enforce the learning through repetition will need to be reviewed/used/consumed at regular artificial interval instead.

I've had a personal wiki for years, and moved to Obsidian.md two years ago. Content that I find usefull or worth keeping is broken down and rewritten into my wiki.

Here it is important to note that simply copy pasting long form text from a source does not work as well as writing the content in your own style. Writing it down, and having to reason about the content as I am doing so helps me retain it for longer.

I then have a list of content that I review at regular intervals (weekly, monthly, quarterly, yearly, 3/5 years). These are simply links within my wiki.

The tricky part here is to not overload your future self. Sure I can write down pages and pages of information and just expect future me to consume it. But in reality I know there is a limit.

I like this approach. I'm likely to overload future me though. How do you determine where the limit is?
It is a balancing act, and it changes over time.

What I considered important to review at X interval last year, is perhaps not that important this year and so I lengthen the time between reviews (simply move the link).

And at some point the content is firmly embedded and does not need regular review.

As others have noted, Spaced Repetition (I use Readwise, Anki is good too) is a good starting point. My workflow for trying to learn new concepts looks like this:

1. Have a habit to review 10 flashcards in Readwise [0] every day 2. As I read, stash snippets of interesting things into Readwise 3. If there are bigger insights I want to digest, I take notes using Dendron [1] and link those insights into other things I'm studying

Eventually, I want to integrate my markdown notes in Dendron with Readwise so that it's easy to block out something I want to review with Readwise in my notes, and for that to automatically show up in Readwise.

[0] https://readwise.io [1] https://dendron.so

(comment deleted)
I study by writing a summary of the material that I'm studying. If the material is very comprehensive I do a summary of the summary. The shortest summary is the only material that I review before the exam. For example, if I had to study a whole book, I would write a long summary first, and then a second, shorter one. This technique works because first you have to go through all the material and understand it, and second everything you write down tends to be remembered much better than if you just read it.