In Praise of Memorization (opens in new tab)

Disclaimer: Anecdata.

I volunteer tutor high school kids in my local community. Particularly math and history. The biggest differentiator between the students who scale up quickly and those who scale up a bit more slowly is their aptitude for memorization.

Recent history curricula and pedagogy in HS has placed an emphasis on “analysis” and “critical thinking.” But the students who struggle all struggle for the same reason: they cannot remember at the moment of writing their paper the basic facts that give context or even give meaning to the documents they are to analyze. I can teach a kid how to do the ACT of analysis to a great level of competency but their papers will be filled with bad analysis and illogical conclusions because they do not know what they read. So, much to their protests, I train them to memorize. And that is often the last push they need to not only get good grades, but realize the joy of critical thinking itself. For now they are not wasting brain cycles trying to conjure up the facts.

Good memorization skills are a huge leverage and it is something that can be learned and practiced. It is a shame that schools are letting children let their mental muscles atrophy like this.

Memorization might be a part of it, but imho the important part is (emphasis mine):

> So they ended up reading *reference manuals* and writing down or memorizing the answers to their questions because they couldn't look up information very easily.

Good reference manuals are dope and they're so much better than tutorials or FAQ like stackoverflow because they are written to be general, you don't have to generalize yourself (which usually isn't trivial). You're almost never reading about a particular problem but about a class of problems, not about a technic but about a class of technics. In freshman we learnt ocaml (actually caml light, yeah you guessed i'm french) and we were all handed a text copy of https://caml.inria.fr/pub/distrib/caml-light-0.74/cl74refman.... To this day i still love reference books, usually when i have a problem i look it up, get my answer, and then end up reading the whole chapter or so giving much more background knowledge.

I've never regarded memorization highly. Whenever we had to learn the times tables or the squares at school, I would just work them out in my head instead of memorizing them.

Of course, I still ended up learning the squares and the times tables by heart, but not because I actively memorized them, but because I just used them so much that I couldn't help but remember them eventually.

I'm of the opinion that this leads to a good rule of thumb: never memorize anything - if you use the thing often enough, you can't help but memorize it anyway.

Of course, you could argue that how often you use something isn't necessarily equivalent to how much utility you might get out of memorizing said thing, and I don't disagree with that.

All that being said, I do agree with the article's premise that an expansive knowledge base aids reasoning, which does seem to be in conflict with my principle. I definitely do possess a basic knowledge of geography, and it does definitely aid my reasoning, but I don't ever remember actively memorizing that - not at school, nor elsewhere.

When encountering a new concept (a definition, a theorem) memorization is more important than understanding. Yes, you can spend time trying to crack the meaning, but you won’t unless you spend enough time contemplating examples and trying to imagine counter-examples. It’s a struggle. A more efficient way of gaining understanding is, first, to have things you do not (yet) fully understand memorized to the letter, and then use this to do exercises and solve problems. Only then you can more or less fully appreciate the concept, its raison d'être, and why it is formulated the way it is; no amount of explaining on the part of the instructor can be as helpful as your own practice actually using the thing.

This reminds me a brilliant article [1] by Barbara Oakley, who has a _superb_ course on Coursera on "Learning How to Learn" [2].

In this article, Barbara talks about how memorization helped her with Mathematics, a subject that she had previously struggled with. In particular, this line stands out to me

> Continually focusing on understanding itself actually gets in the way.

[1] https://nautil.us/how-i-rewired-my-brain-to-become-fluent-in... [2]https://www.coursera.org/learn/learning-how-to-learn

P.S: Her book is really good too https://www.amazon.com/Mind-Numbers-Science-Flunked-Algebra/...

[Update: Apparently she wrote a book based on the course—https://www.amazon.com/Learning-How-Learn-Spending-Studying/...]

This is something I miss from the era when I learned to code: thick books of reference material, designed to be comprehensive. At the time, your compiler was called Superlative Language Number, and you had confidence that as long as you double checked the Number.Subnumber supplement, it would keep working the way the thick stack of paper promised it would.

What you memorize here is the table of contents, and approximately where each answer is inside the book. Search might be better at first but it doesn't scale like logarithmic search over a static pile of paper.

There are programs I can still apply this approach to, vim comes to mind, but for the most part documentation is ambient now, with the leading search engine delivering meaningfully worse results year over year.

In University I had a prof that gave us 600 facts at the start of the year. All printed and easy to read. Each class we went over a few of them.

He said if you mastered 100% you'd get 100% on the exam. He would pick from the lot and that's the exam question.

His reasoning was. These are the basics, we were all smart, and we could use the basics to go beyond them. But only if we had them.

20+ years later and I still remember most of the things in that class. I can still write code in assembly (8086). I don't need to, but I have the knowledge, and it's helped shape my code.

Anyone who's interested in this way of learning, I'd recommend you to check out retrieval-based learning (1):

"Retrieval is the key process for understanding learning and for promoting learning, yet retrieval is not often granted the central role it deserves. Learning is typically identified with the encoding or construction of knowledge, and retrieval is considered merely the assessment of learning that occurred in a prior experience. The retrieval-based learning perspective outlined here is grounded in the fact that all expressions of knowledge involve retrieval and depend on the retrieval cues available in a given context. Further, every time a person retrieves knowledge, that knowledge is changed, because retrieving knowledge improves one’s ability to retrieve it again in the future. Practicing retrieval does not merely produce rote, transient learning; it produces meaningful, long-term learning. Yet retrieval practice is a tool many students lack metacognitive awareness of and do not use as often as they should. Active retrieval is an effective but undervalued strategy for promoting meaningful learning."

1: https://sci-hub.hkvisa.net/10.1177/0963721412443552

Memorizing is a lost art. It breeds understanding, because the symbolic surface makes itself apparent to the mind so readily—and without understanding, the symbolic surface has no grounding.

Only once the lines are memorized can you say them with feeling. And, when you know the feeling, you can make up new lines.

Understanding and memorization should not be seen as opposites.

Memorization also forces you to "reduce dimensionality" in the data—find the underlying patterns of correlation and causality that cause the world to be structured the way it is. You can't feasibly fit every single event in the history of every country in the world into your head, but it becomes a lot more feasible once you extract a narrative that explains the data effectively.

Memorization implicitly requires learning and characterizing them entirely separately as modern education does is pretty faulty.

I largely agree with the author. Even the quickest internet search is many orders of magnitude slower than recollection from memory. The analogy with retrieving from cache vs RAM / disk / network is probably well known. But we should not go overboard in either direction. Everyone's memory is limited, and there are appropriate uses for both calculations and lookup tables.

Quoth Sherlock Holmes:

"I consider that a man's brain originally is like a little empty attic, and you have to stock it with such furniture as you choose. A fool takes in all the lumber of every sort that he comes across, so that the knowledge which might be useful to him gets crowded out, or at best is jumbled up with a lot of other things, so that he has difficulty laying his hands upon it. Now the skillful workman is very careful indeed as to what he takes into his brain-attic. He will have nothing but the tools which may help him in doing his work, but of these he has a large assortment, and all in the most perfect order. It is a mistake to think that that little room has elastic walls and can distend to any extent. Depend upon it there comes a time when for every addition of knowledge you forget something that you knew before. It is of the highest importance, therefore, not to have useless facts elbowing out the useful ones."

Memorization is a stepping stone on the way to mastery.

Reminds me a bit of https://en.m.wikipedia.org/wiki/Shuhari

The agile people have gotten ahold of this principle, but it's pretty cool.

Paraphrased:

Shu - learn to repeat and mimic exactly what the master is doing

Ha - Once you've (memorized?) learned the movements, then you can innovate and add your own twist to it

Ri - Finally, once you've learned to mimic the master and have learned to push boundaries, you can start redrawing them and improvising

There was an HN poster a few months ago who claimed that memorizing the Python standard library made him much more productive.

I memorized half of If when I was younger. It seemed especially impressive to certain people.

These days I combine mnemonics with memory palace journeys. In a few hours of practice I went from being able to memorize 6 digits to around 70.

My penultimate challenge is to memorize texts verbatim. I haven't been able to find a straightforward method. My next thought is to research theater performers' methods and Hafiz methods. If anyone has any specific pointers to share I'd be grateful.

I've worked with devs who never looked up anything on the internet and tried to do everything from memory and their code was shit. I think here it may have been something else that made this team highly productive.

Maybe they had more autonomy than most dev teams are allowed to have. I know I would have never been allowed to create a compiler for any of my previous jobs (to high risk).

I do agree you shouldn't just Google every problem you have.

I wonder if there should be a distinction between memorization and intuition. The author alludes to the fact that if you can build an intuition to the fact that daylight savings time was right in the middle of world War 1, then you can remember more easily that it was in 1916.

I have found that I cannot remember certain mathematical properties at all. For instance I always forget whether to use sin or cos to get the x component or y component. But, I had a few teachers force us to memorize the fundamentals: SOH, CAH, TOA. Using the fundamentals allows me to prove to myself very quickly what I should use. It's the same thing with the quadratic formula. I can never remember it, but I had a calculus teacher show us how the equation is derived. I can remember what a parabola is, and I also remember the fundamentals of calculus. Using these two things I can derive the quadratic formula pretty easily.

So I wonder if there's a 2nd metric that we're missing. Where we need to memorize fundamentals, and critically think about how to use those fundamentals. I don't know, this could just be more anecdata though :)

You know those fancy electronic notebooks? They're barely better than actual notebooks. They all seek to emulate the pen and paper experience, they go to great expense and make it a focus of their advertising that they feel like real pen and paper. Why wouldn't I just use pen and paper?

I believe electronic notebooks haven't yet realized their greatest potential, spaced repetition. I want to take my notes in an electronic notebook, then review my notes and extract parts of them to become spaced repetition items. This is something I would happily have a dedicated device for, and would pay $1000+ for such a device, but they don't yet exist.

After taking notes, natural hand made notes, I imagine creating spaced repetition "flashcards" would work like so:

1. Review notes and draw a square around something I would like to memorize; perhaps a fact I have written, or a hand drawn chart or diagram. This square becomes a "flashcard".

2. Optionally, I may hide (by blurring, or pixelating, etc) part of the flashcard.

When the time came for study, these flashcards would be presented to me following a spaced repetition algorithm:

3. Present the flashcard, including all blurring and pixelation.

4. When ready, the user is shown the full unobscured flashcard. The user may also toggle the surrounding page the square was originally extracted from, this gives the user context.

5. The user can then judge how well they remember the flashcard. The user presses a button indicating how well they remembered the flashcard. This feedback goes back into the spaced repetition algorithm.

The important innovation here is that users can easily combine hand-written and hand-drawn notes and a spaced repetition algorithm. In Anki, you can import pictures, but this is burdensome. You can write markdown or LaTeX, but those suck compared to just drawing something by hand. Studies have shown that hand drawn notes aid in memorization. The would has never seen hand drawn notes easily combined with spaced repetition, and I believe it would be better than any other memorizing scheme we've yet encountered.

Convenience kills.

The Art of Memory by F.A. Yates is still a nice read.

In the article, the engineers admired by the author were intel ops who were stuck in an airgap and had to use their minds. That's a principle I try to use in life - to some extent deliberately making things hard for myself. As with all exercise, what was impossibly hard becomes commonplace.

Part of this discipline is knowing what not to do (yes, attic theory). I never create new accounts on anything or unless I've a clear cost-benefit rationale that makes memorising a new 32 character passphrase worthwhile. That excludes a huge heal of crap that wants you to "sign up" for it.

Or as we call it,in infant school "rote learning" which incurs a huge penalty in popular opinion, despite driving recall of basic arithmetical facts like the times-tables, alphabet order, spelling "rules" &c

Understanding that memorization should be key for the reasons laid out in the article, would it be better to forgo note centered workflows such as Zettelkasten or Wiki for flashcard centered workflows such as Anki or Supermemo? What gives the most value?

I'm going back and forth between these two workflows and I feel like progressing at a steady rate before getting bogged down, not because I have difficulty with the subjects I study per se, but because every month or so there is an article confuting the previous one, prompting me to switch to the other workflow, slowing me down considerably, every single time.

I think heuristics are much more important than knowing semantics. Someone who can memorize all the notes on a piano faster than their peer isn't necessarily going to be better at piano in the long run.

Similarly, there was an idea expressed in chess where some people rely on memory to reproduce thousands of similar scenarios, while others rely on the "feel" of the chessboard and the various heuristics related to it (e.g. is the king "safe" -- e.g. castled and not exposed etc., are the bishops "powerful and free to roam" -- e.g. not blocked by pawns or heavily contested etc.).

My argument partially comes from self-defense; my memory is comparatively poor. But I'm able to use heuristics developed over years of experience to make generally good decisions.

So in this case:

"Who would you rather hire: the person who knows exactly what features are available in PHP 7 and which are only available in PHP 8, or the one who will figure it out by trial-and-error of while writing each application and seeing what fails?"

I would hire the person who is able to write the best application, and understand when to use what features or not: similar to the Bond quote "Sometimes a trigger needs to be pulled -- Or not pulled, it's tough to know which in your pajamas". If memory was the most important part of cognitive ability computers would serve a much larger role in society than the (already large) part they do today.

> Subconsciously, when you learn a piece by heart, its message penetrates deep inside you. It lies at your fingertips, ready for you to make use of it. Many cultures have long understood this. In Islam, people who memorize the entire Koran are given the special title of hafiz, or guardian. In a secular equivalent, I know people who have memorized Rudyard Kipling's poem If— to give them a moral helping hand at times of crisis.

What poems you all think serve as good reminders of truths of how to work well in teams and build good software?

Honestly the ancient Latin's dict "est modus in rebus" (there is a measure in anything) IMVHO is still generally valid, witch means that yeas we also need memorization but there is no reason to stretch it like in the old time before printed press. I fail to found a purpose to remember exact date of birth and death of historical figures, remember poetry etc

Similarly I see no point in doing the opposite, like certain "GTD fanatic" who pretend not to remember what they need to do in next few hours because their GTD system is more efficient than their memory and so they need to keep memory free to focus on current thing.

Est modus in rebus.

We need to eat, eating too little is harmful, but also eating too much is harmful. We need oxygen to breathe but too much exigence fry our lungs and brain. Sport is important, a sedentary life is not healthy, but too much physical activity means being broken in the old age with a consumed body.

The most important thing to operate with their own brain IMVHO is having in memory "the big picture" so to being able to recall anything via computer as needed. Without the big outline our searches are limited, probably biased, we can't correlate things we encounter etc, but there is no need to do more and no reasons to do less.

Use method of loci. It's an almost trivially learned skill, and once you've got the technique down, you can expand it as needed, or riff on the basics with other mnemonics to amplify the speed and volume of memorization capacity.

Rote learning is difficult, and unnecessary. Your brain already automatically memorized places, faces, and novel things. If you can imagine Elvis in a wizard robe holding a block of cheese on your front porch, you have a one item memory palace. That's the level of difficulty a memory palace requires - it's a hack of built in automatic functions your brain is already performing.

Elite memory athletes who memorize the order of thousands of cards, or tens of thousands of digits of pi, or a travel dictionary, or any other assortment of things - these people are often average intelligence, otherwise normal folks who just practice a neat trick.

There are dozens, if not hundreds of "advanced" memory, mnemonics, or method of loci primers. Get one and develop a superpower.

I like "Memory Improvement" by Ron White - you can grok it in an afternoon, but it's structured into 30 daily exercises, and captures all the important features without fluff or filler. The audiobooks is great, 4 hours long. A motivated person could master the skill over a weekend with nothing but the audio.

Seriously. Method of loci is a basic human skill that gets attributed to genius characters like Sherlock or Tony Stark or Einstein level intellects, or to shamans and druids and mystics. It's a default feature that comes in every install - it's how your brain wants to work. Every single person can do it, and should.

This analysis is only relevant for neurotypicals. There are ways that your brain can work that flip this whole thing on its head, and if you have one of those brains you need to do exactly the opposite of what this article recommends: You can only remember the details by understanding the context. I do not understand that Oman is powerful because it controls the Strait of Hormuz; I do not understand that Russia wants Crimea because it has Mediterranean access. Rather, it's only possible for me to remember that Crimea has Mediterranean access by understanding that Russia wants Mediterranean access and wants Crimea and using those facts to infer the relevant geography. I am able to remember that Oman controls the Strait of Hormuz because I understand that Oman is powerful and that the Persian Gulf is important and I can work backwards from there. So be careful about the directionality of causation when you're looking at observations like "The most understanding-capable people I know know a ton of details" - they may be able to know details because they understand rather than being able to understand because they know details.

Memorization alone and in itself is nothing but just a curiosity. But coupled together with good analytic and connection skills it is an incredibly powerful asset.

Very much in agreement with the article. The analysis about the current Ukrainian situation is a little bit limited, though, because Russia controls Sevastopol since it annexed Crimea in 2014; it didn't need the current war to control it more.

Also, probable typo:

> A second example: at the time of the writing of this article, Russia is a month into an invitation into Ukraine.

invitation => invasion? I don't think anyone invited Russia in Ukraine.

> I once worked at a small company of insanely productive engineers.

Which one? https://www.linkedin.com/in/pearlleff/ only shows internships at Tumblr, Amazon and LinkedIn.

This sounds like those contrived morality tales that people write on LinkedIn to prove their point.

For math, when I was in school I tried memorizing all the formulae, and well, my life was miserable. I hated it.

Recently just for fun (I have a job that (perhaps unfortunately) does not require any advanced math), I started taking math MOOCs from top universities. I discovered that, when I looked into how the formulae is derived (starting from elementary math), I find it substantially easier to keep in mind the formulae. Or even if I don't remember the formulae, I know how to derive them.

I now feel like I've understood certain concepts much better, even though if there's a timed test I might end up being worse off than others who memorized. But then again, I'm learning for fun and is not preparing for any test :)

What I'm trying to say is, for math atleast understanding the underlying concepts work better for me, even if at a given point it's beyond your grade.

(Thank goodness MOOCs exist. I should probably donate.)

I get the impression that memorization fell out of favor in education because in the past, it was used as an ends in itself rather than a tool to generate insight.

In school, my dad had to memorize a lot of dates with precision down to the day. That seems mostly-useless to me. Just knowing the decade is sufficient for most analysis. To take the article's example, the putative link between "Row v. Wade (1973) → Downswing in crime since the 1990s" only really requires that you know that the ruling occurred sometime in the 70s.

I think the strategy, then, should be to memorize things that are likely to generate insights and to the level of precision that will likely be necessary to do so. You should probably commit to memory that Andrew Johnson was the president after Lincoln in 1860s -- but you probably don't need to remember that Millard Fillmore succeeded Zachary Taylor on July 9, 1850.

Memorization is one of God's greatest graces to the intellectual.

One of the big advantages of memorization is that you can work on problems in your sleep.

Sometimes when I have a very difficult problem, I will memorize all the information related to the problem and then go to sleep. Many times, I have woken ip in the middle of the night with a solution to the problem.

> “An overly precise memory is maybe not really what we want in the long term, because it prevents us from using our memories to generalize them to new situations,” she said in San Diego at a recent meeting of the Society for Neuroscience. “If our memories are too precise and overfitted, then we can’t actually use them to … make predictions about future situations.”

- https://knowablemagazine.org/article/mind/2019/why-we-forget

I guess it depends on what is meant by "memorization." I feel like I have a good memory but I suck at memorizing things. I would say I have knowledge but when it comes to dates and quotes I am terrible. I can remember things that happened and my friends and I usually do well at pub trivia. However, in school there were two times we were supposed to memorize poems and I sucked at it. I screwed it up both times. When we would take a test and afterwards people would ask what I got in question 4, I would have no idea.

>Many would argue that there is no point for kids to memorize the world map today.

Al Franken can draw the United States map from memory! (He claims it was for a bar bet.)

Al Franken and Tom Davis visit Late Night with David Letterman, 8/20/87. Al draws a map of all 48 contiguous states in under 2 minutes.

https://www.youtube.com/watch?v=Mn2ofGwDd4A

While I agree that factual data points are necessary to test the hypothesis and reach a better conclusion, I disagree that one needs to memorize them. Facts are stored in databases and books. One can always look it up. Maybe a matter of minutes or hours.

There is no point in memorizing a readily available fact that virtually has no effect on your life.

"Subconsciously, when you learn a piece by heart, its message penetrates deep inside you. It lies at your fingertips, ready for you to make use of it. Many cultures have long understood this. In Islam, people who memorize the entire Koran are given the special title of hafiz, or guardian. "

Mentioning Muslims who learn by heart the Quran as a means to illustrate how memorization makes the message penetrating deep inside us, is something totally wrong. VERY wrong.

There is no single Muslim in the world (and in the history of Islam, including Muhamad himself) who knows the meaning of Quran. One of the obvious reasons you may understand is that a large amount of the Quran vocabulary is not Arabic (even the word "Quran" itself, is not Arabic). Basically you can pick up any random Quranic text and you will not find one single Muslim in the world who knows at least the meaning of certain words. If you read the Islamic resources which explain the Quran, you will find dozens, and sometimes hundreds of explanations for the same text just because some words are not even Arabic. So memorizing the Quran by heart does not help you in anyway to understand its message.

I could have said similar things about other examples you mentioned. But the point of my comment is that between programmers we should stick to programming concepts when we try to explain something instead of picking examples from fields which are beyond our knowledge and which approach is merely a waste of time and confusing at best.

Also do not try to overthink a concept. It's easy to be quickly off-topic.

I still believe one the biggest problems new students have at community college from their math classes is lack of memorized basic math. When you have to take any time to do the basic math when learning college math, you are at a time disadvantage. Flash cards are apparently the devil, but you don't need to work out the simple stuff and calculators are context switches.

Understanding is easy, memorizing is hard.

The book "A Mind For Numbers" by Barbara Oakley highly praises memorization as a way of learning.

And from a personal perspective, it (memorization) has proved highly effective for my learning.

Like my 7th grade Geography teacher used to say, "Repetition is the key to memory". I remember that because he made us repeat it a bunch of times.

Just a heads up for the author, if they are reading this thread at some point:

> Russia is a month into an invitation into Ukraine.

Probably should read 'invasion'.

Yeah it's a huge differentiator. But there's limits and tradeoffs. Likely those guys were super productive in one very narrow thing.

Your neurons in your brain made up the CPU.

your memory is the DDR and cache, we all know the more the better.

muscle memory is cache, it has the fastest access.

Good note