OP and others here are stretching the definition of “memorize” to mean “anything that leads to something being retained in memory.” I reject this idea.
The trauma of burning your hand on a hot pan creates a memory you won’t soon forget, but almost no one would understand it as an act of memorization.
Memorization to me refers to a set of cargo-culty “learning” practices wherein we believe that by using language to drill exposure to an abstract representation of a concept, that somehow we will absorb the concept itself.
We do this mainly because experts suck at empathizing with learners and fail to understand that the symbol has meaning for them but not for the learner.
It’s the difference between drilling vocabulary flashcards and actually reading, listening, or talking to someone.
Young children do not use vocab flashcards to learn their L1. They aren’t being “drilled” to learn “mama.” They have actual needs in an actual social context and attend to nuanced details of that context to make complex statistical inferences about the world, their perceptions, and their body. Mostly subconsciously.
Yes, there are specific areas where drilling can help us accelerate or catch up. Many kids seem to need explicit phonetics instruction in order to make the leap to reading words. Phonological speech interventions are often drill-like. Practicing musical scales does make you more fluent in improvisation. Drilling the mechanics of a repertoire piece frees your mind to focus on higher-order expression and interpretation. They’re valuable, they have a place.
But this is just a small slice of learning. It’s disproportionately important for passing tests (And getting hired at tech companies!), which to me is the crux of the issue.
If I had to reformulate OP’s argument to something I can get behind, it would be more about deliberate practice or “putting in the reps.” This is also often boring, and differentiates highly successful people from average performers. But it’s a broader and more purposeful set of activities than “memorization” would imply.
> We do this mainly because experts suck at empathizing with learners and fail to understand that the symbol has meaning for them but not for the learner.
I hear you; but teaching deep expertise is really hard. We can use your example of a child learning their first language. They will really understand it. But people are famously, hilariously terrible at teaching their native tongue. We know how to conjugate, and how to use verbs and adverbs and all the rest. But it’s all intuitive - we have no symbolic understanding of it. If that’s the case, we can’t explain it in words.
Here’s a weird fact: if you look around the room you’re in now, I bet you know what it would feel like on your tongue to lick everything you see. We probably learned that in the “put everything in your mouth” baby phase.
You are an expert. But if you wanted to, how would you teach that? I think the learner would just have to go lick a lot of things for themself.
I believe a lot of real learning is actually like that. When I taught programming, I think I was a frustrating teacher. My students would ask things like “what’s the best way to structure this program?” And I would say “I don’t know. Let’s brainstorm a few ways then you should pick at least one and try writing it like that. Figure it out in code.” I think you become great at programming by licking all the programs you can find. Same with music and art and languages (go have conversations with native speakers).
There is only so much the best teacher can teach. Sometimes you just have to walk around licking things.
> There is only so much the best teacher can teach. Sometimes you just have to walk around licking things.
There's being taught and then there's exploring and practice. It's the exploring and practice that is often missing today, IMO because we all want everything to happen so quickly.
In my sport of choice, jiujitsu, there's been a shift happening in teaching. Traditionally the teacher would show a move or two then have the class drill the move. The shift now is more conceptual where the teacher might say the goal is get chest to chest and then have the students operate at 20% figuring it out on their own. While there are some pretty vocal camps on both sides, I think having both methods is the best to learn from. Being taught a solution to a problem a student discovered on their own during discovery is when learning really takes hold IMO.
Finally, the best way to really learn that I have found is to teach something.
> I hear you; but teaching deep expertise is really hard.
No it's not; this is the point of apprenticeship. It's just not compatible with the educational institutions people actually pay for.
> I think the learner would just have to go lick a lot of things for themself.
This is not contradictory. If your idea of teaching is just lecturing at people and occasionally verifying specific knowledge, obviously they're not going to learn very much. Your job as a teacher is to facilitate a growth curve of increasingly more difficult and unwieldy problems to solve.
> Here’s a weird fact: if you look around the room you’re in now, I bet you know what it would feel like on your tongue to lick everything you see. We probably learned that in the “put everything in your mouth” baby phase.
You are an expert. But if you wanted to, how would you teach that? I think the learner would just have to go lick a lot of things for themself.
This is such a weird and amazing analogy. Thank you for sharing it! I don't know if I've ever heard this concept explained in quite this way before.
I quite liked your licking metaphor/symbolism. In fact, if I find out that something tastes good after licking it, I'll go even further. I swallow it. Ha ha.
My intuition is that creativity is a more of a feeling one has, rather than the result of some teaching method or cultural background. It seems very closely related to one's interests and also the novelty effect. Just think about the first time you experienced something, like a great song or your first love. The mind is just bubbling with thoughts.
How can that feeling be experienced again with the same object like it was the first time? It can't. It's never the same as the first time. Even if we can mentally forget, the subconscious somehow remembers it's had that experience, so it isn't fresh anymore.
So, yes, you need to make connections between existing things to come up with something new, but once the fascination with a thing dissipates and it becomes familiar, any further creativity in that area quickly dies off. Our interest just moves on.
Perhaps the really creative people are just the ones who have this ability to see familiar things with ever fresh eyes?
Kinesthetic learners know exactly what you are talking about. Others may have no clue. Just a moment ago I identified which pill bottle I needed to open by shaking it. The sound prompted an image of the pill shape and density. This interconnectedness of sign systems and our human ability to read them can truly emerge in brilliant every day ways. I agree nothing replaces experience, especially for kinesthetic learners.
All those really learning a foreign language get to the point, where they "chug it" and jump into the deep end, surrounding themselves with native speakers and going to that learning by context mode. Its the toughest phase, but works.
The EEs I have known that carried around a card with:
V = I * R
I = V / R
R = V / I
because they couldn't remember it were all bad at EE and bad at math.
If you can't remember the pieces making up a concept, how are you going to remember the concept?
> It’s disproportionately important for passing tests (And getting hired at tech companies!)
I don't remember anyone who couldn't pass tests but was really a great engineer.
BTW, one of the tests fighter pilots go through is they are blindfolded, and then have to put their hand on each control the instructor calls out.
I also have some written tests for certifying pilots. There are questions like max takeoff weight, fuel burn rate, max dive speeds, etc. Stuff a pilot had better know or he's a dead pilot.
> I don't remember anyone who couldn't pass tests but was really a great engineer.
But there are plenty of people who can pass tests and are terrible engineers. When people talk about memorization they talk about remembering the words without understanding the concepts, such a person is no better off than the person carrying around a text note with the concept written down.
When you understand these concepts you will remember those formulas, but memorizing the formula doesn't make you understand it. Therefore creativity doesn't fundamentally comes from memorization, there is something else there.
> I don't remember anyone who couldn't pass tests but was really a great engineer.
This is just confirmation bias. Either you're in a field where you must have a degree, so everyone who couldn't pass tests simply never became an engineer (eg EE) and whether or not the test is a good measure is irrelevant because it nonetheless gates your sample
OR you're in software engineering and the people who struggled on tests work beside you, but they don't tell you about their past performance on tests because you have a chip on your shoulder and they don't want you to look down on them.
I failed a lot of tests in college and now I'm a great software engineer.
Some times the brain is wired to intuit the concept. That's something that fascinates me. You grasp the idea before any articulate explanation. Somebody shows you a problem and rapidly you start discussing solutions with the other person and even go further.
Most of the time memorization is a key role for creativity, the easier you can jump between ideas the more combinations you can explore (seems like the brain is constrained by some cache bottleneck in a way).
> If you can't remember the pieces making up a concept, how are you going to remember the concept?
This is not mere memorization. The GP's point that not everything that makes you remember this is simply "memorization", and certainly all of those things contribute differently to creativity.
> I also have some written tests for certifying pilots.
Yes, piloting is one of the tasks where you absolutely need lots of random information memorized. You should also not do a lot of "creativity" in it.
Isn't that exactly the result of a focus on memorization instead of understanding (yes yes at some level that also involves memory), though? To a certain type of person memorizing an arbitrary arrangement of 3 symbols is quite difficult but OTOH it's easy to remember that current goes up with voltage and down with resistance.
Zero of the EEs I have known need that (where did yours graduate from?)
The mnemonic to remember V = I * R is as a triangle with V at the top and I, R on the second row, the same way you remember the relation Distance = Speed * Time or Mass = Density * Volume in high school. Or any other triangular multiplicative relation.
This is different memories? the one for the formula and the one for the concept. So you can remember the meaning of the word but don't remember how to write it. So can reproduce the formula from your deep understanding, but it is quicker to check it out instead of apply the first principle every time.
Saddest part is that you only need first one and and then with memorization of most basic algebra you get the others.. Not even need to do it symbolically.
> because they couldn't remember it were all bad at EE and bad at math.
> If you can't remember the pieces making up a concept, how are you going to remember the concept?
You touch upon the different levels of knowing it. Yes, having to carry a card with the formulas on it shows no knowledge. But, if you have to memorize the formulas, your knowledge is still not adequate. You're just regurgitating a formula that you memorized so you can plug in numbers. You don't understand the "why" of Ohm's law. Of course voltage is equal to current times resistance, it's obvious by what these things are! It should be as self-evident as "Of course distance is equal to speed times time!"
Another example: You can have the Lorentz transformation formulas memorized but still not really understand the "why" of Special Relativity.
My high school had majors and one of them was effectively EE; the 14/15 year olds in my major had those formulas down inside of 3 weeks... what EEs did you know that couldn't outperform teenagers?
...I mean, I didn't even go to that great of a college and no one would have made it past the second year of EE without memorizing 10x more formulas than that.
I'm just completely lost on how it's even possible to have an EE degree and needing a card. Signal processing classes required math 100x more difficult than that. I had to know quaternions by my second dsp class.
Memorize means "to retain in and quickly recall from memory". Weather that is by synthetic or natural process is irrelevant. From the point of understanding how memory and recall work, yes burning your hand is an act of memorization.
Sure there is a natural repetitive process that leads to base learning like L1 you mention.
On the other hand no one adds or multiples enough in daily life for natural memory formation. Humans consider the skill vital enough that we have developed methods to memorize them. Same for spelling, especially for infrequently used words.
Flashcards used with Spaced repetition isn't cargo cult, it is a well studied, and pretty good method of inducing memory formation.
I’d say that’s the definition of “remember” rather than “memorize.”
To most people I’d wager “memorize” has a strong connotation for the synthetic version only, with an emphasis on a stripping-out of context.
I recognize that stripping away context can be valuable—-drilling a tennis serve over and over outside the real-time context of a game is extremely helpful.
Flashcards are rarely valuable in the same way. For semantically oriented tasks, an impoverished context is usually not very helpful. Receptive skills like letter and character recognition might be an exception. But even then you’ve got to make the leap to reading at some point.
> no one adds or multiplies enough in daily life for natural memory formation
To the contrary, there’s a fascinating study of children in South America who had very fluent mental math skills for making change because they sold fruit on the side of the road. They couldn’t solve the exact same problem in story problem format in a classroom, though. Synthetic contexts usually don’t transfer well to real life.
By far the best way to learn arithmetic facts is to ‘naturally’ use them in service of solving more interesting or relevant problems. Someone who spends the same amount of time doing nontrivial word problems, pattern-discovery projects, playing a game or solving a puzzle involving embedded arithmetic, or just talking about numbers in a group will come out vastly better prepared both for recalling or figuring out arithmetic solutions per se and for mathematical fluency in general than someone who does narrow practice drills. Arithmetic drills are not only a total motivation killer for most people, but also just suck at aiding retention. Time spent on arithmetic drills in school is somewhere between a waste of time and an actively harmful punishment.
> Memorize means "to retain in and quickly recall from memory". Weather that is by synthetic or natural process is irrelevant. From the point of understanding how memory and recall work, yes burning your hand is an act of memorization.
I think in common usage the word "memorize" very strongly implies that it's the lossless storage and retrieval of some highly specified sequence of information. No amount of studying American Civil War history would be referred to as "memorizing the Gettysburg Address" unless you could recite the speech word for word.
> On the other hand no one adds or multiples enough in daily life for natural memory formation. Humans consider the skill vital enough that we have developed methods to memorize them.
Isn't that kind of the point for the other side?
Nobody memorizes the answer to 212+457. There are no flash cards for every possible addition up to infinity.
In my experience, the more people think that math is something to memorize, the worse they are at math.
>OP and others here are stretching the definition of “memorize” to mean “anything that leads to something being retained in memory.”
They "stretch" it to its dictionary definition?
>The trauma of burning your hand on a hot pan creates a memory you won’t soon forget, but almost no one would understand it as an act of memorization
It still is a kind of memorization, just not a voluntary one. And such learning is still is a very important function of mental development and evolutionary fitness, that shouldn't just be shunned "because trauma".
>Memorization to me refers to a set of cargo-culty “learning” practices wherein we believe that by using language to drill exposure to an abstract representation of a concept, that somehow we will absorb the concept itself
Well, I see your cargo cult and I raise you tried-and-true.
"Absorbing the concept itself" might take more effort (including personal, for logistical reasons, not everything can be tailored to the individual learner, who might not even care enough for learning compared to all kinds of diversions, and have zero passion for the subjects, even if a clone with the teaching skills of Feynman with the presentation skills of Tonny Robbins, and the passionate conviction of Jean D' Arc was to present it to them.
But absorbing the concept is not enough, there needs to be instant recall, or at least fast enough recall) of all kinds of facts and factoids and tables, and also this "absorbing" also needs to encompass boring concepts, that are nonetheless crucial, if one is to be succesful in anything technical or scientific, or generally creative in any sort of organized way that combines concepts and information (not just Jackson Pollocking away).
To memorize means to commit to memory. It is an action. But memories are created by not just actions, but experiences. Experiences can create memories without you having committed them; without you having memorized anything.
They actually are, thou. Small kids are repeated the same things over and over, hundreds of times. They watch the same Bluey episodes dozens of times, get them read the same books dozens of times, listen to the same songs dozens of times.
> It’s the difference between drilling vocabulary flashcards and actually reading, listening, or talking to someone.
Those are not opposite activities. Drilling vocabulary flashcard is the most efficient way to start being able to read/listen/speak; and it's not even clear from research that output (speaking and writing) is useful at all for learning.
Also good luck learning to read Chinese or Japanese without rote learning a few hundred characters. Even native speakers learn them by repetition. You can't be serious advocating "just go read stuff" as a way to learn a language. Some foundation, acquired by explicit learning, is required to even start reading.
> Young children do not use vocab flashcards to learn their L1.
That's a bad argument: the life of a toddler is vastly different than an adult. A young child has basically nothing but figuring out what's going around him (including language) at least 16 hours per day, every day. An adult has much less time for that.
> Drilling vocabulary flashcard is the most efficient way to start being able to read/listen/speak
This isn't actually a settled matter. I did a literature dive a while back and found that drilling vocabulary flashcards shows the highest benefit on artificial recall tasks (like multiple choice tests), and over relatively short time scales (days to weeks). Studies that looked at longer time scales (months to years) and more organic tasks generally showed mixed results. Which I generally interpret as a sign that the literature in question is highly susceptible to the file drawer effect.
And that in turn suggests that the magnitude of flashcarding's value for this kind of thing has a lot to do with your goals. In a nutshell, are you more interested on the science on how to develop communicative proficiency over the long run, or are you more interested on the science on how to get a good grade in class?
I'm studying Chinese right now, and I do use flashcards, but it's not because I believe it's the best way, per se. It's because it's a convenient option for reviewing characters and words that don't appear often in the reading material I have. When they do appear often in my reading material, I find (anecdotally) that it takes a lot fewer organic repetitions to get the character or word to stick in my head than it does with flashcard repetitions.
It's also worth mentioning there's no particular reason to assume Chinese and Japanese schools are any less likely than schools elsewhere in the world to cling to inefficient pedagogical techniques out of a sense of tradition. So one can't necessarily assume that the way they are doing it is the way they ought to be doing it.
> That's a bad argument: the life of a toddler is vastly different than an adult.
Also it takes a surprisingly long time for children to learn language. I used to think it sorta-kinda happened over a year or two, but having children myself revealed how wrong I was.
I have written up a fairly hefty dictionary of words they mispronounce or even just invent on their own because they don't know the one in their native tongue. My oldest rarely contributes individual words to this dictionary anymore, but he still improvises expressions and idioms.
> and it's not even clear from research that output (speaking and writing) is useful at all for learning.
Only if your goals don’t include being able to speak or write.
> You can’t be serious advocating “just go read stuff”
True, not “just.” And Chinese is particularly tricky because the ideograms convey little to no phonetic information. Even so, almost any activity I can imagine seems superior to traditional flashcards. Photo flashcards (vs. translation), listening along to highlighted text or closed captions, deciphering street signs or memes, even the written drills you mentioned. (Or better, “write 5 phrases that all start with character X”). Our brains crave meaning, and flashcards offer very little of it.
> even native speakers learn them by repetition
No argument there, most everything is learned by repetition, but I’m interested in context. Native speakers already know the verbal form of most words they’re learning to write, even in Chinese. I’d argue the meaning is stronger.
> life of a toddler is vastly different than an adult
True. The scale of their learning tasks are much bigger than ours. They have to learn that they exist, that their family exists, that they can vocalize, that language is a thing, that they want and need things, and that they can get them by communicating.
> An adult has much less time for that
I think this is a good entry point to the core of the issue for me—-small children don’t “set aside time to learn,” they just learn. You and I do this also, though it’s less novel and flexible and therefore maybe less salient. I think we place too much value on structured or synthetic learning as “real” learning when in fact it’s often extremely inefficient compared to our natural learning tendencies.
There’s a spectrum of structure, starting with what we choose to pay attention to, to an open-ended “study time,” to guided classroom activities, to timed math drills. Flashcards are at the extreme reductivist end of that spectrum. I suspect we like them because they’re easy to understand, uniform, predictable, and convenient to create and use. Creating more effective learning opportunities and supports is substantially harder, but generally worthwhile IMO.
Vocabulary flashcards are not all that efficient way to start being able to read/listen/speak. They teach you translation rather then meaning directly, you don't get context or the "context" is super repetitive sentence and so on.
And plus, general recommendation is to learn words elsewhere and just put them into anki to not forget.
Some people like it, but it is not the only or the most recommended way to learn speak and write.
> Also good luck learning to read Chinese or Japanese without rote learning a few hundred characters. Even native speakers learn them by repetition.
I've tried this for years with Japanese kanji and never really got very far. Just didn't work well, they largely were just a big blob of lines.
Then I found an Android app (Kanji Study) that mixes this in with informational screens that break down kanji into radicals and puts them alongside a bunch of multi-kanji words and uses them in sentences so we can see them in context, and it's actually been working.
You are changing the definition of memory we have had for like 25 centuries in the West. That is your choice but there is no stretching on the other side.
> The trauma of burning your hand on a hot pan creates a memory you won’t soon forget, but almost no one would understand it as an act of memorization.
I reject you rejection. You need to provide a reason something learned via pain is fundamentally different from something learned without pain. We used to beat school children for not learning their lesson.
Drilling vocabulary flashcards is very helpful in learning a language. As is reading, listening, or talking to someone. To learn a language well, you need to do both. Although children ultimately reach the level of being a native speaker, they do not do this very quickly considering that they say their first words when they are about 1 year old.
haha, no man, the "creativity" cargo cult should be entirely ignored. You shouldn't even consider what mechanistic formula they've invented. If you do you might buy into it and your creativity will be less your own.
Never listen to the grownups, everything they say is a lie.
I am not commenting on or agreeing with the OP, but your response is false. LLMs aren't given just a dictionary, and they do not know how to speak. Speech implies grasp of semantics. There is zero semantics in a block of text, only, according to some interpretation, various textual correlations.
> It’s the difference between drilling vocabulary flashcards and actually reading, listening, or talking to someone.
You need the 'flashcards' before you can read, listen or talk. Go try reading a book where you don't know most of the words. Heck you need 'flashcards' before you needs 'flashcards for words'. You need to memorize the alphabet first. Try reading a text where you haven't learned the writing system.
> Young children do not use vocab flashcards to learn their L1.
Because they can't read.
> They aren’t being “drilled” to learn “mama.”
Obviously you aren't a parent. You think a child magically decides one day to say mama? Or do you think it's the mother constantly saying 'mama' to the child until the child 'remembers it' and repeats it?
> They have actual needs in an actual social context and attend to nuanced details of that context to make complex statistical inferences about the world, their perceptions, and their body.
What? Complex statistical inferences about the world?
> Heck you need 'flashcards' before you needs 'flashcards for words'. You need to memorize the alphabet first.
I've a toddler who can read 3 paragraphs of 3 sentences each, and then tell you the details of the story he read[1]. He is 4y6m, right now. He has never learned the alphabet or the names of the letters (A,B, C, D, etc). He has only learned the sounds a letter or sequence of letters make for certain patterns.
You most definitely do not need to memorise the alphabet in order to learn to read!
Teaching children the alphabet before teaching them reading makes it a lot harder for them to learn actual reading.
[1] I've seen kids as old as 7 get confused by a book with no pictures, and he sails right on through because I taught him to read (using the DISTAR alphabet), and made sure none of our daily lessons had even a single picture in it.
Or maybe we just don't want to coddle them. When has learning anything been easy, and why do you expect people to be able to acquire new knowledge and skills without putting in the effort? It shouldn't be grueling, not for its own sake, but yeah, you might have to stare at a compiler error for a few hours or even a few days before you figure out what's broken. Truly, how else are you supposed to learn if you don't, eventually, do it yourself?
I'm so sick of this anti-expert, anti-knowledge attitude. It's why we have bootcamp juniors being thrown into otherwise-senior roles, with laughably predictable consequences for the field.
Perhaps I worded it too emotionally. I mean that experts struggle to remember what it was like before they understood something. It’s very common for experts to ask novices to make leaps that they aren’t capable of making, because they seem natural or obvious from an expert POV.
I’m all for hard work; learning is usually very hard work. I also think we need expert guidance.
But let’s make the difficulty useful/effective rather than counterproductive.
This attitude seems unrelated to the topic at hand quite frankly. Experts suck at empathizing with learners not because of this spite, but often because it's actually quite difficult to switch gears in language and understanding. It's a completely different way of sharing knowledge, where you have to explicitly express things that are just assumed shared understanding among colleagues.
Also, to answer your questions in a very simple way: the entire reason you even became an expert is because another expert somewhere along the way gave you an easy in to the knowledge, they coddled you. This is why you can "stare at a compiler error for a few hours before figuring out what's broken". Without that expert, you wouldn't even understand what a compiler even is.
"a flash of inspiration connecting internalized concepts"
Well, okay, but rote memorization is neither necessary nor sufficient to internalize concepts.
One of the reasons people make fun of the author's approach to creativity is that systematic memorization fundamentally can't teach taste—so the systematic approach reeks of awkward, try-hard, low-brow, tasteless art.
More broadly, memorization doesn't help much with any sort of tacit knowledge, not just taste. I just figure taste is especially important in creative endeavors. That's definitely the case for programming! Memorization in programming gives us architecture astronauts and design-pattern soup rather than elegant code.
For what it's worth, I do think that it is useful and important to have a good mental model of what expertise is and how you can develop it. Memorization might be a component of this, but it's going to be a small component at most. I expect that realistic practice with fast feedback and expert mentorship matters far more. (If you're curious, I found the book Sources of Power by Gary Klein gave me a good way to think about how expertise works.)
At the same time, memorization has a real cost: it takes time and it's frightfully dull. For me, at least, trying to memorize something without context is not just ineffective but also totally kills any intrinsic motivation I have for whatever I'm learning. Sometimes a bit of memorization is unavoidable, but I've found that to be relatively rare. Otherwise, my time is generally better spent on some sort of practice in context.
One of the points I'm trying to make is that taste and elegance fundamentally stem from an internalized heuristic -- which at it's core is memorization.
I understand the connotation of "memorization" evokes an image of blindly memorizing without connecting, but isn't the tastefully developed expertise just memorization of a better heuristic?
I don't think I can agree, as an extremely creative person with extremely bad memory - to a point where I pretty much never memorize anything, whether intentionally or by accident.
What I find instead, is that by just processing novel information, especially if I focus on analysing it, my brain internalizes insights and builds model of that type of thing, allowing me to either imperfectly reconstruct what I've seen, or to come up with an infinite array of permutations, extrapolations, etc which is where the real ideas come from.
Further, ideas crucially revolve not around just the information itself, but the "feel" for what role they play in the whole, how well they do it, in what way they're notable, etc.
In fact I'd straight up claim that memorization is antithetical to creativity - a perfect ML autoencoder or GAN would just regurgitate the training data. Creativity comes from generalisation while memorisation is analogous to overfitting.
That's true if you broaden the definition of "memorization" to cover all learning, but "learning is necessary for creativity" would not be a particularly interesting thesis.
Expertise is the result of learning from past experience, both in developing an internal intuition for what you're doing and in having past patterns to draw upon. To the extent that experts have simple easily verbalizable heuristics, these are largely post-hoc attempts at explaining their intuition rather than an accurate reflection of how they make decisions.
And, in fact, experts can't even always do that: it is perfectly possible for experts to make good decisions without being consciously aware of why they are making them, and explaining how to make good decisions is a separate skill from being able to make them in the first place. The book I mentioned has a memorable story about a firefighter who thought he had precognition after pulling his team out of a dangerous situation without any specific indicator of the danger, but I figure a more common example is experts saying they did something because it was the "obvious" or "clean" or "better" way to do it and getting a bit flustered when pushed further.
We can see this in action pretty clearly if we look at advice for, say, writing. There is a lot of advice from good writers but just memorizing and blindly following this advice is actively counterproductive. Advice you can memorize fundamentally must lack nuance and context. We can see this clearly because so many different pieces of writing advice contradict each other and because good writers do not follow any of those suggestions with any consistency.
The same definitely applies to programming, which is why we have both "don't repeat yourself" and "you ain't going to need it", and why new programmers trying to apply either rule (or both!) to a codebase inevitably create a mess. What I've found with programming advice is that most suggestions are either actively wrong or too vague to be useful. (By the time you've learned enough about programming to be able to follow the vague advice, you don't need it very much!)
Why are you attached to the word "memorization" here? Certainly taste comes from experience and learning. Maybe you could argue that all learning is an oblique and imperfect form of memorization—but why argue that at all?
The only reason I can see is if you think memorization could be a shortcut to good taste, which it can't. Acquiring good taste requires broad experience—more information than you can possibly remember—such that you retain a suite of sophisticated intuitions. Cutting that information down to something that can be memorized would require you to (1) already have the intuitions you're seeking to acquire, and (2) be able to express them all in plain English, which, as far as I know, cannot be done. No painter has ever expressed their aesthetic in such a way that a student could memorize that expression and then have the same creative sensibilities as the original painter.
Ultimately, there's no substitute for the process of simply consuming lots of art while paying close attention to what you like about it.
>One of the points I'm trying to make is that taste and elegance fundamentally stem from an internalized heuristic -- which at it's core is memorization.
seems to me there is a relatively big inductive gap there, you believe that there is an internalized heuristic and at its core is memorization, you may even have some evidence that this internalized heuristic has strongly informed your development of taste, but it is pretty difficult to make an argument that is the case for all people.
Aside from that I would say that "internalized heuristic with memorization as the core" puts everything on nurture and no input of nature - which I am pretty much in the camp of combinations of nature and nurture creating the person - of which taste must surely be a big component.
not the parent poster but I think I agree with your perspective here. The alternative is that some individuals' taste or sense of aesthetics is somehow innate and unmoored from the statistics of the things they experience. There may be something to this, but for most practical purposes I would agree with your point.
also a question - if you have better long term or short term memory how does that affect taste? How does it affect creativity, if all of these things are essentially memorization you would have to assume that people were more creative and had better taste the greater their ability to memorize things, which in the case of taste especially seems slightly absurd.
In the case of creativity it may be easier to make an argument - but surely you can find people who seem more creative with less ability to memorize.
> Memorization in programming gives us architecture astronauts and design-pattern soup rather than elegant code.
Elegance is probably orthogonal to creativity, and likely follows from some kind of minimization principle, like minimum program length. You are effectively distilling the "essence" of something from all of the noise.
Creativity seems different, more like novelty, and creativity following some kind of remix of memorized elements + some randomization seems very plausible.
You can create something novel but not elegant, and something elegant but not novel, and you can distill an elegant version of something novel that your or someone else created and that's the best of all creations.
I completely disagree with your assertion that "...rote memorization is neither necessary nor sufficient to internalize concepts."
I would recommend reading the book Moonwalking with Einstein. There is a lot of discussion there on how memory is linked directly to creativity, and to understanding concepts deeply.
---
A choice passage:
"...If the essence of creativity is linking disparate facts and ideas, then the more facility you have making associations, and the more facts and ideas you have at your disposal, the better you'll be at coming up with new ideas. As Buzan likes to point out, Mnemosyne, the goddess of memory, was the mother of the Muses.
The notion that memory and creativity are two sides of the same coin sounds counterintuitive. Remembering and creativity seem like opposite, not complementary, processes. But the idea that they are one and the same is actually quite old, and was once even taken for granted. The Latin root 'inventio' is the basis of two words in our modern English vocabulary: inventory and invention. And to a mind trained in the art of memory, those two ideas were closely linked. Invention was a product of inventorying. Where do new ideas come from if not some alchemical blending of old ideas? In order to invent, one first needed a proper inventory, a bank of existing ideas to draw on. Not just an inventory, but an indexed inventory. One needed a way of finding just the right piece of information at just the right moment.
This is what the art of memory was ultimately most useful for. It was not merely a tool for recording but also a tool of invention and composition. "The realization that composing depended on a wellfurnished and securely available memory formed the basis of rhetorical education in antiquity," writes Mary Carruthers. Brains were as organized as modern filing cabinets, with important facts, quotations, and ideas stuffed into neat mnemonic cubbyholes, where they would never go missing, and where they could be recombined and strung together on the fly. The goal of training one's memory was to develop the capacity to leap from topic to topic and make new connections between old ideas. "As an art, memory was most importantly associated in the Middles Ages with composition, not simply with retention," argues Carruthers. "Those who practiced the crafts of memory used them---as all crafts are used---to make new things: prayers, meditations, sermons, pictures, hymns, stories, and poems." ..."
Great book, motivated me to then read The Art of Memory by Frances Yates.
Although I'd say traditional mnemonic devices like memory palaces are basically linear information storage and recall devices. This can create issues in building a flexible web of information, because loci or the order of the path can become dependencies and you can run out of unique spots in a given space, leading to memory interference.
Even spaced repetition methods (e.g. Anki) tend towards fragmentation of micro-ideas. Its perfect for terms, languages, and simple one question -> one answer ideas.
I've found a hybrid method of images, nested loci and spaced repetition to be most useful, because its flexible over time, and preserves relationships of ideas.
(Context: I co-founded a SaaS in this space: www.sticky.study)
The book Make it Stick taught me that this Don’t Cramp My Style With Your Boring Rote Learning (Man) attitude is prevalent in teaching. At least American. They argue that it is wrong for the same reason that the author does.
But saying this to a programming crowd must be the most futile thing. At least instrumentalists have to rote train their muscle memory. That lowest bar has to be passed, even if it’s just three chords.
But the article isn’t about programming creativity though. It is a general concept. But if honing in on the mythical lone-genius activity (geniuses never practice in a structured way) helps you win an argument then so be it.
I think it is more of becoming fluent with primitives that can be composed in versatile ways. I can see how that can be poorly understood as memorization.
The main implication is that if what you are “memorizing” is not easily composable, then you won’t be able to apply them broadly or creatively.
However, I disagree with the author on what creativity is, although his definition is one experience of a creative inspiration.
Interesting perspective. I do agree that there are people out there who develop a distinct "taste", but I can't tell if this refers to a "style", an emergent property of multiple "habits", etc? I've always wondered how one develops their "taste".
Also, would you consider a subconscious habit "memory"? What's the difference between the two?
> One of the reasons people make fun of the author's approach to creativity is that systematic memorization fundamentally can't teach taste—so the systematic approach reeks of awkward, try-hard, low-brow, tasteless art.
Well... can you think of an artist who didn't have a deep knowledge of their art-form before they pushed it forward? Three that jump out for me, in no particular order, are Picasso, Borges and Jack White. After all, great artists steal.
There's a big difference between "artist who didn't have a deep knowledge of their art-form" and "artist who didn't follow an explicit system to memorize a bunch of rules to make their art".
Wait, why do you think Picasso didn't have deep knowledge? He studied both at the School of Fine Arts in Barcelona & the Royal Academy of San Fernando in Madrid, for ~5 years before moving to Paris.
Borges was incredibly talented, but it's worth keeping in mind his dad was a writer too.
Good art very much relies on being exposed to lots of other good art first. I don't know that rote memorization is the best way to achieve that, but you definitely need that exposure.
I think that argument is that these artists did not memorized rules or previous pictures and then applied them. They did put a lot of effort into learning, but that is different claim. If you define "memorization" as "any learning of anything", then the word is kind of useless.
> Well, okay, but rote memorization is neither necessary nor sufficient to internalize concepts.
Of course it is. It's how every human child learns initially. By rote memorization. How does a toddler learn how to say mama? By constantly hearing and repeating it. How does a kid learn their ABCs? Rote memorization is the basis of all memory.
> Memorization in programming gives us architecture astronauts and design-pattern soup rather than elegant code.
Dumbest thing I've ever read. You write programs well by doing and remembering. Same with writing. Memorization is the necessary component to programming well. In other words, you program well by remembering elegant code.
> For me, at least, trying to memorize something without context
After the basics, most memorization is contextual.
> At the same time, memorization has a real cost: it takes time and it's frightfully dull.
Oh dear. Something isn't fun all the time. What a childish worldview. It's more fun to eat candy and drink soda than eating 'dull'. It's more fun to sit and watch youtube than to workout.
> Sometimes a bit of memorization is unavoidable, but I've found that to be relatively rare.
Relatively rare? In order to be competent in anything, you have to memorize lots. You can't write a good essay without having memorized much of the material. Trying reading a book where you have to constantly look up definitions of words because you lack the vocabulary. Try having a conversation with someone who has to constantly look up words because he lacks the vocabulary. Try having code review with someone who doesn't remember anything about their code.
> Otherwise, my time is generally better spent on some sort of practice in context.
Why? Because it helps you remember?
To the idiot ( probably OP ) who downvoted, try coding without having 'memorized' the keyboard. The anti-intellectual, anti-hard work, anti-memorization agenda pushed by some 'people' online bears looking into.
> Growing up with Indian parents in California, I was exposed to both. My mom would write daily Kumon sheets out by hand for me to do, and teach me from Indian textbooks from the same grade (which were much more advanced than the US equivalents). The result was me breezing through the US school system without much thought.
Ukrainian refugees I know are finding the same things in the UK school system, where the maths is much less advanced. Philippines schools, meanwhile, have better discipline and more motivated students.
I conclude that Western public education is in a bad state, and this is a source of chronic social weakness.
With the caveat that I haven't looked in a while, the rot seemed very anglo-specific. Coming over from Germany almost three decades ago, it was amazing how much US text books just didn't cover.
There were a few developments in Germany that pointed in the same direction, but there a large gap at the time. Meanwhile, even back then, eastern Europe certainly had even higher standards. (I replaced a lot of math textbooks with a copy of Bronshtein & Semendyayev)
The influence of religious and conservative types has weakened US textbooks. Do a deep dive into how Texas can dictate what is presented in schoolbooks nationwide.
Maybe because UK education system knows about this PISA and Ukraine and other countries don't even care?
For example one thing that puzzles me in western education is this reading comprehension. What exactly is this for? I studied in Russian school/university and we had nothing like that.
Reading means you should understand what you are reading, reading without comprehension is just nonsense for me.
Also, this problem is a vicious cycle. Bad math students become bad math teachers. I believe that one has to be significantly good at a subject in order to teach the basics well. In school I got a feeling that many math teachers did not know a chapter of math beyond what was in the textbook.
Doesn't help that in the American economy, good mathematicians can earn 10x more than a math teacher's salary.
Indian math system is not memorisation based either.
Math anywhere in the world is all about how many patterns you know/seen. And inventing new math is all about changing one small thing about the stuff you already know and see if it remains logically consistent.
Knowing what to change largely depends on the stuff you already know. To that end even if you don't memorize facts, you still need to memorize patterns.
You have to memorize something. Nobody invents the Apple pie from scratch.
I was the sort of person who did not believe in memorisation as a solution for anything. Then I tried getting really good at spaced repetition for a year (yes, it is a skill that needs to be trained for good results) and I've completely changed my mind.
Spaced repetition allows me to become proficient even in things I don't get the natural opportunity to practise daily, so that when the day comes and I need them, I have some level of knowledge already. This has happened to Kubernetes troubleshooting, statistics, PowerShell windows programming, and traffic engineering just in recent history.
I have yet to publish some of these, but I have examples from statistics:
I'm similar. This from your top link stood out to me:
"It’s a little like building with lego bricks or something – spaced repetition helps ensure all the tiny pieces are in the right place, so that the big castle can happen without structural integrity issues."
The book Make it Stick (by Henry L. Roediger III) had a similar idea they called 'Structure Building'. Very similar to what you described, more experienced and effective learners were creating mental schemas of how the little, but crucial parts of a subject fit together, and successfully cut through the noise.
Structure Building was associated with interleaved practice (shuffling of problem types) and spaced retrieval practice.
I really enjoyed both blog posts, thank you for sharing! And I have to say, your explanation of the subexponential distribution property was remarkably clear for someone without a background in statistics :)
Would you mind sharing the flashcards you generated to build this intuition? I've been using Anki for a while and really trying to focus now on improving my prompt writing; would love to see how you managed it for this problem.
As much as I would like to, I think getting to that understanding required at least 500 flashcards on general statistical and probability concepts, ranging from fundamentals to extreme value theory. Most of those are only barely relevant at face value, but still contribute to understanding.
It's not that I set out to understand this specific thing but that I had studied statistics with flashcard support for a year and that happened to work after a few attempts.
This is similar to a loose life-thought i've had for a while, though i lack a catchy phrase for it lol.
My thought is: You're always practicing a behavior whether you like it or not.
Your mind is always setting you up to do more of whatever it is you're doing now. Both physical and mental. To do it more efficiently. With more ease. With more frequency. etc.
It's good motivation for me to mitigate a lot of negative behaviors. Angry in traffic, self anger, etc. If i can reason at least, of course. As i'm not interested in doing a lot of things more than i am, so i should avoid doing them now - if possible.
Math proof and derivations are a bit like remembering a walking route. You've seen the start and end, and the main turns taken, and there's also a general "walking" skill you need.
My argument is that it is worth memorising also the derivations, rather than re-deriving from scratch each time.
Meorising the derivation makes it easier to derive a second-order derivation, and so on. At some level of abstraction, going from first principles becomes prohibitively expensive and caching intermediary results, or so to speak, unlocks that again.
I think memorization gets a bad rep because you need to be acutely aware of what you're memorizing, like memorizing the sequence of an answer sheet instead of core concepts. But when done sufficiently rigorously, the foundations of memorization make room for higher-level critical thinking and reasoning.
Practice is an oft suggested solution to developing mastery, but I did like how the article framed it: creating subconscious heuristics and memory.
Absolutely, especially in real world application. If you don't have the ability to pull on fundamental ideas anywhere, anytime, then have you really mastered the learning material?
Right. I noticed this acutely in an abstract algebra course. We learned several different proof methods, then the exam was just "prove these theorems with the tools you have". I'd never been challenged with math like that before. I mean, I bombed it lol, but nobody was going to pass if they didn't remember, say, how to do a proof by induction or what it means. At some point, you need to be able to recall this information. Maybe the psychologists categorize these things differently, but I'd argue it's clear that some form of memorization is necessary for the task.
It’s just a bunch of arbitrary unprovable assertions.
Everyone here seems to have, broadly speaking; neither a) the qualifications to knowledgeably comment of the (honestly poorly understood, afaik) function of “creativity” or b) anything more meaningful than “here is my naive personal lived experience and opinion” to contribute on the topic.
It’s just armchair psychology.
If you want to wax philosophical, by all means, but I think anyone taking “thoughtful insight” away from this article or thread is fooling themselves.
Whatever the epistemic quality of the article is, it's triggered some interesting discussion here which I think is valuable. No need to denigrate talking about human experience with other humans, I think?
You'll experience the creativity outlined in the article directly when you start doing deliberate memorization, i.e spaced repetition. No qualification needed.
I think semi-obvious would be a better criticism than unprovable.
You can't make connections unless you have things to connect to.
You can't recognize (your own discovered/inspired) novelty unless you have memorized normality.
If you are creative in the absence of knowledge of what already exists then that's considered as reinventing the wheel, and not very useful, even it it's Ramanujan reinventing much of established mathematics.
One might talk about it from the perspective of birdsong, which is used by mates to judge sexual fitness. First a tutee bird learns from a tutor bird, and then eventually applies variability to the original song.
It's strongly suspected that anterior forebrain pathway (AFP) may be a source of behavioral variability. We naturally age over time, including our vocal musculature, so in some sense we must constantly relearn how to use our muscles to deliver a song.
When a bird is deafened its birdsong will naturally drift, but when we precisely damage the AFP along with deafening we find that birdsong remains stable for a longer period of time, until of course inevitably it must drift due to aging vocal musculature.
It's pseudoscience, author would benefit from reading (and memorizing) the current scientific literature on learning and cognition.
There's a bad trend, Hacker News gets this kind of blog-style self-promotion every month that gets much comment attention, but the essays are not well-researched and with made-up assertions written by programmers talking out of their lane ("Engineer's disease") and not having done homework on the subject.
> It's pseudoscience, author would benefit from reading (and memorizing) the current scientific literature on learning and cognition.
How so? Can you share what you have read and that is relevant and applicable in real life?
> There's a bad trend, Hacker News gets this kind of blog-style self-promotion every month that gets much comment attention, but the essays are not well-researched and with made-up assertions written by programmers talking out of their lane ("Engineer's disease") and not having done homework on the subject.
You are on the wrong site, this is not a scientific journal there is no need for scientific rigor in every post and comment.
P.S. The real life world is full of events and things happening if you cannot learn by yourself (create theories models on how the world works aka. pseudoscience apparently) and need academic verification for everything. Then you some kind of disorder :/ good luck
Alphazero was very creative, yet it didn't memorize a single move, it just self played. Deep blue was not creative at all, but it was the chess engine that memorized the most moves, todays chess engines are much better at chess and they don't memorize many moves at all, if you dig through their internals you wont find a lot of board states there.
So no, creativity doesn't fundamentally comes from memorization, memorization is neither sufficient to become creative nor is it a requirement. You don't memorize concepts you build models around the concepts. You wont be able to reproduce the exact descriptions of concepts but you will be able to produce something similar that means basically the same thing.
All very nice and handwavey, but then you see the user’s current venture is scammy deepfakes as a service, which is about as creatively bankrupt as it gets.
Shame about the national stereotypes as well. There is plenty of creativity in Asian countries. Just bizarre assertions all around.
Historically eastern Asian cultures have placed duty to a whole host of things before oneself, and in many cases the old aphorism "the nail that sticks up gets hammered down" applies as well. Plus Japan and Korea have well established cultures of adherence to tradition and mastering simple, time tested things to a ridiculous degree rather than trying to innovate.
Not true of everyone, but if you compare a culture that values conformity and tradition to a country that values the freedom for the individual and trying new things, of course it's not going to measure up by western standards of creativity.
Have you seen their art and entertainment? I assure you there is whole lot of creativity in there. And it has a whole lot MORE variety then western tend to have.
But LLMs to date can't really differentiate well between a creative insightful answer, and a nonsensical one. The selection process is still done by a human.
you guys might be interested in the latest Machine Learning Street Talk podcast [1] which right from the start is all about how LLMs are great for creativity, as in novel combinations of trained data (from memorization) -- but are not capable of the reasoning skill needed to verify if any idea is actually plausible given a set of constraints.
I have to say, you were naive to ever think that. Rolling a die can produce creativity — and I’m not being facetious.
Creativity, at least in fields traditionally considered to be ‘creative’ like music and visual art, is 90% randomness and 10% retroactively attempting to ascribe meaning to the randomness. And this is coming from someone who hugely values and defines oneself via an ability to be creative.
Also, LLMs haven’t hit ‘escape velocity’; that terminology has spread like a meme (because it’s an attractively dramatic yet easy to understand idea) but isn’t backed up by any science. Maybe one day we’ll have the fabled ‘AGI’, but ChatGPT isn’t it.
This seems to resonate with my experience, although I feel myself bristling due to the baggage of the word memorization.
Although sometimes “memorization” doesn’t happen because you sit down to do it but rather that you keep using the same things over and over when solving problems that they become internalized. I find that to be a more fruitful path towards understanding that I don’t want to call memorization but it is.
Maybe you should first try to separate the negative feelings you have towards the word "memorization" and the word itself "memorization". There's nothing bad about memorization. This sort of negative bias about inconsequential things is something that can easily hold us back from things that could help us further ourselves.
And agreed -- it's this exact realization that led me to both this method and title
Imo this negative connotation has made many people refrain from calling internalization what it is
But acknowledging that it's memorization has actually made me more efficient at learning, since I can now consciously look for the heuristic, codify it, and try to commit it to memory
Kind of shocking to see so much angst against memorization here.
Memory has been long thought to be a critical component of intelligence, with elaborate mnemonics systems developed by people to help memorize more things (see Francis Yates' The Art of Memory, and to a lesser extent her book on Giordano Bruno).
I would contend that memorizing concepts is a first step in understanding them. Also, that generally understanding concepts isn't a one and done thing, usually there's layers. Personally I found that memorizing things in math helped me immensely when years later I needed to actually understand the things I had memorized.
A fellow CS student didn't understand a theorem, because he didn't understand any of the three definitions used to state it. We went
through the definitions together and suddenly the theorem was "trivial"
That is true understanding. He won't need to remember the theorem, because in his mind, it automatically follows from the raw data.
> The inspirational lightning bolt writers and artists experience can't happen unless they know how to write or draw.
It absolutely can happen (from my experience), but you might not be able to do anything about it. Like I have new melodies and songs pop into my head fairly often, but I'm not good enough at making music to translate those ideas into finished songs (I have made some songs with FL Studio in the past, so it's not totally impossible, but my focus has been on other things).
I also have ideas for stories fairly often too, and while I could write them, they tend to get backburnered for my game ideas, which I also have.
For me, I have plenty of creative ideas, I just don't have the energy or system built up to help me get those out there in a fast enough manner, and I likely won't ever have that, it would pretty much require other people to handle almost all the rest of the work beyond the inspiration and testing out a few things.
As an example, I've made over 60 prototypes for board game ideas in the past five years alone, but still have only gotten one of those games picked up by a publisher. If I were Reiner Knizia, all 60 of these games probably would have been released, because he's built up an engine around him (people willing to playtest all of his designs enough to be polished, and has enough of a reputation that finding publishers willing to publish most of his designs is super easy), and never has to worry about the look and feel or the manufacturing or even the theme of the game, that will all be handled by the publishers.
The spontaneous melody inspiration is probably because you're exposed to music a lot and could easily recompose using those scales, instrument concepts, etc.
And 60 prototypes sounds a lot like practice to me. I wonder if that person you named (who I, a layman have never heard of) made a few games before they got to where they are. All that practice builds up ammunition for creativity.
The consensus I'm seeing reading these comments is that the article is a lot more useful if you replace "memorization" with "deliberate practice". The first implies little deep understanding but lots of practice. The second implies both.
I think if you've ever been exposed to music your mind could come up with different melodies, and who hasn't been exposed to some sort of music in their life. Even if they were the only human on the planet, they'd probably still encounter songbirds, at least.
I disagree that you have to be exposed to a lot of music to even have a chance at coming up with something. Maybe coming up with something good, sure, but not having that spark of inspiration. People can have flash inspiration for garbage ideas too, it happens all the time in startups :P
That being said, the spark might happen more often the more you practice doing something about it, to your point.
Reiner Knizia is the most prolific board game designer of all time. He has 742 games credited to him on BoardGameGeek (some of those are rethemes of older games of his, but he still has probably at least 500 different games released).
About half of my prototypes are about on the level of cleverness and completeness as the lower rated half of the games he has published, at least, and maybe even a couple might slip into the top half somewhere (at his most brilliant his games are way, way better than mine though). But his name carries a lot of weight so he doesn't struggle to find publishers for a good chunk of the games he comes up with.
Meanwhile I've mostly switched back to video game development because at least that way I can easily publish the game myself when it's ready (game development is super hard in other ways, though). I tried for six years and got one signed board game that still hasn't been released after four years. Meanwhile there was a time where I was releasing 1-2 Flash games a year. If I had just made video game versions of my game prototypes instead I probably could have some basic playable version of it out on a website somewhere.
Although I do still submit some board game designs to unpublished game design competitions (I was a finalist in one of the biggest ones, for a design which still hasn't found a publisher) and go to conventions and set up meetings with publishers when I can afford to do so (I couldn't afford to this year, my spouse took a break from work and our income cut almost in half).
What's more is that memories are just a replaying of neuron connections activating in the brain - and when we are prompted by the world around us those connections will fire in response to the stimulus. Quite similar to how AI neural networks function - which is why I believe that AI can indeed be creative and create "new" ideas
I actually made a video diving deeper into this and comparing responses from people and ChatGPT for a creative thinking problem - https://youtu.be/l-9EUBbktqw
I think your hypothesis here (and probably the entire article as well) is strongly challenged by the 'progenitor argument.' Take humans at the dawn of humanity. Language did not even exist beyond what may have been crude sounds or gesturing and collective knowledge did not fall that far beyond 'poke him with the pointy side.' Somehow we went from that to putting a man on the Moon in what was essentially the blink of an eye.
Training an LLM on the entirety of knowledge at this dawn of humanity and, even if you give it literally infinite training time, it's never going to go anywhere. It's going to just continue making relatively simple recombinations of its training set until somebody gives it a new training set to remix. This remix-only nature is no different with modern knowledge, but simply extremely obfuscated because there's such a massive base of information, and nobody is aware of anything more than a minuscule fraction of it all.
---
As for the 'secret' of LLMs, I think it's largely that most language is extremely redundant. One thought or point naturally flows.... why do I complete the rest of this statement? You already know exactly what I'm going to say, right? And from that statement the rest of my argument will also mostly write itself. Yet we do write out the rest, which is kind of weird if you think about it. Anyhow the point is that by looking at language 'flow correlations' over huge samples, LLMs can reconstruct and remix arbitrarily long dialogue from even the shortest of initial inputs. And it usually sounds at least reasonable, except when it doesn't and we call it a hallucination, but it's quite a misnomer because the entire process is a hallucination.
I agree with the author, at least in my own creative experiences. However, it's more likely the case that 'creativity' is arrived at differently for everyone. I find memorization to be a comforting foundational activity that builds knowledge & confidence, which I can later express creatively.
There is deliberate practice for skill-building. There is exploratory "making" that fuels originality. There is inspiration hunting and incremental tweaking to get to creative mutation. There is high productivity that triggers eventual ingenuity. I find the article hyperbolic in its thesis and execution especially when it comes to the final hand-wavy bit about how there is more per-capita creativity in non-rote learning.
While its hard to prove or disprove without a long study to prove or disprove the author's claim, I'm willing to die on the following hills:
1. Kumon sheets are the antithesis to creativity
2. Understanding is not a form of memorization (not the rote variety anyway)
I've thought a lot about education, and my personal take is that in the US we way undervalue drilling (by which I mainly mean building up familiarity and muscle memory) and way overvalue understanding.
I've been collecting quotes about these topics for a few years. One relevant to creativity and drilling is Bob Dylan's
> If you sang "John Henry" as many times as me.... you'd have written "How many roads must a man walk down?" too.
There is definitely a lot of value in practice and repetition. I don't think rote memorization / drilling are the only means of getting that practice and repetition. Ironically, with a bit of creativity, we can provide both. Lot of practice, lot of repetition, paired with understanding, play and making things.
Creativity is actually defined by this transformation of finding connections between raw data that you already have to know.
Consider use cases for a rock.
Boring would be using it as a paperweight or throwing it through a window.
Novel but uncreative would be throwing it at the sun, or painting it red. Novel, but kind of useless.
But what about using a rock to play rock paper scissors? Planting it in the soil and watching it grow? That's kind of novel, by way of subverting rock's rules (it doesn't grow, unlike plants) or transforming the concept of 'rock' itself — a real rock isn't needed for rock paper scissors.
So only connections between known concepts are creative. Others might be novel, but useless.
Being forced to do rote exercises sometimes makes you creative. Solve a thousand trivial multiplication problems and you will spontaneously discover lots of shortcuts, patterns, intuitions that can warn you when you make a mistake and so on.
A common issue I notice when people discuss the terrible state of math education in the US is that teachers demand that you solve a problem a specific way, such as multiplying two-digit numbers by drawing base-ten blocks and applying the distributive property.
People who are good at doing multiplication in their head think the method makes perfect sense and don't know what all the fuss is about. But I believe that those people learned how to apply the distributive property "by themselves". That is, by adults forcing them to multiply two-digit numbers over and over until they developed an intuition of the distributive property by necessity.
When people who didn't go through countless drills are taught the base-ten method directly, they have a harder time understanding it. So ironically it is the students who "mindlessly" drill trivial computations over and over that are more prepared to have a "true" understanding of the distributive property, while the ones whose teachers believe drilling is for chumps and try to just explicitly show them the true distributive right away, they end up memorizing the words of the distributive property without understanding it.
I think this analysis as a bit guilty of over-fitting -- it is quite easy to rote memorize a bunch of things while having little to no understanding of what they are or how they work. Trivial examples include training a room of people to memorize a series of facts written in a language they don't speak (the fact that they have memorized doesn't at all mean they have any understanding of the content). So I would say it's not memorization per say, but meaningful exposure to a thing, the more chances you have to meaningfully interact with a thing, the higher the chance is you will learn how to manipulate it and do things with it. This is the difference between understanding and mere memorization, and the more exposure you have, the greater the chance you will start to see the patterns and understand, versus focusing your efforts on memorizing which will just lead to over-fitting and not understanding. As with NNs, so with humans.
On the other hand, episodic memory (insofar as that is distinct from "3*9=27" memorization) is built on top of creativity.
The vast majority of what we consider "memories" are the creative brain doing an on-the-fly story generation, massaged until it "seems right" and serviced plus a big dollop of emotional confidence.
I hear with llm/chatGPT people don't have to blog anymore, but rather the generated plausible-sounding well-structured nonsense flows like an open sewer onto the web.
1. Creativity in a commercial context once stolen/cloned through back-channels accrues value, and manifests as several competitive campaigns
2. New disruptive ideas are usually shelved until the IP/patents expire. No one wants to go through the sometimes impossible licensing process
3. Emerging technology is usually degraded in the rush for IP assets by established firms i.e. large firms dump billions on ridiculous concepts out of fear of market fragmentation
4. Startups do not usually have cash to burn on speculative IP. Thus, real cutting-edge experimental technology is sometimes never made public for numerous reasons.
I disagree with elements of this article, but enjoyed reading it. True creatives do not subvert norms consciously or with an acute awareness as part of this article suggests, I agree they need to be exposed to the norms to generate their own interpretation, but I don’t believe that true creativity is a conscious exercise.
Well said - what's funny is that many creative ideas (from an outside perspective) are often very simple connections of two existing ideas for the person that actually did something creative! It's just that other people don't have the same context/knowledge as the creator, so an idea can seem extremely original to them
You have to know things to be reasonably creative but there is a point where memorizing more stifles creativity. Memorization is often very passive and is fundamentally different from searching for new ways of doing things. It's hard to make sweeping statements like this because there are different modes of memory and different modes of creativity.
There is a lot of memory involved in being creative, but I think setting out to memorize things is a bad way to be creative. You have to practice being creative. In doing so, you will naturally remember a lot of stuff like what works, what doesn't, and most importantly which types of things you ought to memorize. For example if you're programming you will find it useful to remember the syntax of your languages. If you're writing you'll find it useful to remember styles and vocabulary. And so on...
Good discussion on this article. I went to a Vampire Weekend concert last night and they used the 45 minute encore to play covers that were requested by the audience. They got through about 10 songs (a verse or two and chorus), everything from Talking Heads, to Creed, Sublime, Beastie Boys, and Prince and Bob Dylan (the concert was in Minneapolis, so they probably rehearsed the last two). So basically the band knew/memorized the tune for hundreds/thousands of songs from the past couple decades and the singer knew/memorized the words. So I think that this example supports the authors premise that creativity comes from some form of memorization.
Regarding Asian cultures and memorization, there is a recent change that has to be pointed out. The role of memorization classically was to aid manana/contemplation after shravana/hearing-the-teaching. The stage of manana is a kind of inquiry where you raise questions and clarify them or see how the teaching applies to various situations in your life(Like learning a principle in physics and solving problems based on the principle). In the case of poetic works the inner-feeling/bhaava sinks into you allowing one to access blissful states.
However, in the modern context, this has been transformed into memorizing arcane lists/tables, on which one is graded without any further manana.
That thesis is way, too oversimplified and a bit misleading. It could lead you to think the real source of creativity is how much you store, facts or patterns. Put Wikipedia on a computer with a pile of heuristics (eg How to Solve It). Then, it will be more creative than humans in no time. Yet, humans with virtually no knowledge are more creative.
My first study of creativity was Cracking Creativity by Michalko. Skimming its tools, I noticed all of them were about changing how one looks at a problem or connects its pieces. Another work told me geniuses are defined by what they forget or ignore more than what they memorize. The two ideas combined into my working approach.
When I did CompSci research, I would look through the papers for the concepts they reported on. High level ones, core prerequisites, and techniques in how they are combined. From there, I could enumerate variations on each. Then, just keep combining them in straight-forward or random ways. Eventually, something emerges.
Likewise, prior work on creativity and A.I. showed reasoning by analogy was huge in humans. We find patterns in one domain that we generalize to look for patterns in another domain. Then, there’s some process of knowing when to try one or not.
These processes so far are extracting a tiny amount of information, filtering most of it, identifying heuristics in a domain, and heuristics across domains. Also, letting the mind just soak on things to do whatever it does in the middle of the night. These are collectively creativity.
Memorization is a building block of, but different from, creativity. The proof is how we’ve long had memory and reasoning in systems but they sucked at creativity. Recently, systems are extrapolating enough to be more creative but are hallucinating nonsense they definitely didn’t remember. So, they’re orthogonal.
“Creativity comes to those who have internalized the patterns of their art -- they can see the connection or novelty because it's all in their head.”
Edit: I’ll also add that, for planning and creativity, many of us use a deliberate, trial-and-error process that takes time. We don’t just “see” it from something we internalized. We work toward it using the creative process. We usually do see it when it’s finished, though.
The acts of the mind, where in it exerts its power over simple ideas, are chiefly these three:
1. Combining several simple ideas into one compound one, and thus all complex ideas are made.
2. The second is bringing two ideas, whether simple or complex, together, and setting them by one another so as to take a view of them at once, without uniting them into one, by which it gets all its ideas of relations.
3. The third is separating them from all other ideas that accompany them in their real existence: this is called abstraction, and thus all its general ideas are made.
John Locke, An Essay Concerning Human Understanding (1690)
It’s not memorizing, it’s actually knowing and understanding the utility of different concepts. When you learn of a problem in a new field, you go through your bucket of tools and modify one to fit the new problem.
Probably you would have had more agreement, if you would have used a different word (though I’m not able to suggest that different word that captures the meaning I think you intended)
I'm not in anyway an expert, so I googled what some research says. Here's an interesting meta-analysis (https://link.springer.com/article/10.3758/s13423-023-02303-4). Memory and creativity are a lot more complex than I realized. There are different types of each, and it seems like they interact in complex ways. Here's the findings from the abstract:
> We found a small but significant (r = .19) correlation between memory and creative cognition. Among semantic, episodic, working, and short-term memory, all correlations were significant, but semantic memory – particularly verbal fluency, the ability to strategically retrieve information from long-term memory – was found to drive this relationship. Further, working memory capacity was found to be more strongly related to convergent than divergent creative thinking. We also found that within visual creativity, the relationship with visual memory was greater than that of verbal memory, but within verbal creativity, the relationship with verbal memory was greater than that of visual memory. Finally, the memory- creativity correlation was larger for children compared to young adults despite no impact of age on the overall effect size. These results yield three key conclusions: (1) semantic memory supports both verbal and nonverbal creative thinking, (2) working memory supports convergent creative thinking, and (3) the cognitive control of memory is central to performance on creative thinking tasks.
So some memory seems to be correlated with convergent creativity, which according to wikipedia (https://en.wikipedia.org/wiki/Convergent_thinking) is "the ability to give the 'correct' answer to questions that do not require novel ideas, for instance on standardized multiple-choice tests for intelligence." It sounds like there's less correlation with divergent creativity, which (again from wikipedia (https://en.wikipedia.org/wiki/Divergent_thinking)) is "a thought process used to generate creative ideas by exploring many possible solutions."
But my real takeaway is that people here seem to have strong (emotional?) opinions on "memorization vs creativity: which is better", but few people seemed to bother reading page 1 google results on the topic. So I like to think that bothering to do some cursory research beats both. :)
I see it like this: after I have seen some colors and a ball I can then imagine a ball in any of those colors.
In other words, everything I have experienced and memorized becomes this pool of resources I can imagine from. More I have seen, more combinations I can imagine.
Then I understand how traveling actually broadens my view. It’s not just some nice phrase but hard reality.
Also this means maybe anything we can imagine we can also create. Because whatever I can imagine I can also plot a path from here to there, imagine all the steps in-between.
You may as well say that creativity comes from writing. Because obviously all of the most creative writers write. And the writings of creative non-writers are entirely absent.
Just because you can put things into boxes does not mean that everything belongs in a box. Whatever essential element you seek to create, whether or not it is concrete or abstract, can simply be put forward as a target of memorisation, without pausing to think about whether you can truly memorise it.
E.g. a heuristic for determining the best heuristic. Simple, just memorise it, right?
"Creativity is just connecting things. When you ask creative people how they did something, they feel a little guilty because they didn't really do it, they just saw something. It seemed obvious to them after a while. That's because they were able to connect experiences they've had and synthesize new things." - Steve Jobs
If you are transforming the knowledge, i dont think that is the same as memorizing it. In fact, i think most people would describe that as the opposite of memorization.
Using your definition, everything would be memorization. For example, you could describe picasso as simply pattern matching on other paintings to make something new, but i think it would be crazy to describe his work as an act of memorization (then again, the way AI generated art works... maybe its not so crazy)
> But how do you know how to apply this acquired knowledge in this new context?
That question has a false precondition baked in. If you know how, it's not creative.
> It's some form of pattern matching right -- which imo is just a less obvious form of memorization
No. Sensing and matching patterns does not imply memorization. Everything you're saying is completely loaded.
Did you just discover memorization? Because the pattern I see in your words is similar to anyone who's just learned a new tool or technique - they overapply it everywhere as they learn to use it.
To me, “memorization” implies an active process focused on learning a particular set of “matches” (to adopt the term you’re using here). But it seems to me that tacit knowledge (and other products of less concentrated/deliberate learning) often plays a substantial role in creativity.
That is, creativity fundamentally comes from internalized knowledge (as the article says) but internalized knowledge doesn’t necessarily come from memorization.
I think this is right, and one implication is that 99.9% of the self-help content you passively consume on the internet will have no impact on your life, because (barring deliberate learning / repetition), you won't remember it in the high-leverage moments when it might have made a difference.
> you won't remember it in the high-leverage moments when it might have made a difference
But how would you learn to remember the right advice in the right context? The act of remembering something when you need it requires you to do more than just memorize it, you have to properly learn when the thing is actually useful and train your mind to recall this advice in those situations.
It is much better to know that an advice exist and learn when to look it up than to memorize the advice without the ability to realize when you would need it.
This brings up an interesting point -- while I think many people read and forget self-help books without ever improving their lives, the way they can mostly help people (imo) is by:
1) identifying a heuristic
2) making memorization easier through stories
The stories, analogies, acronyms they teach all just make memorizing/remembering/learning their heuristic easier
This claim makes little sense because it fails to distinguish between memory and memorization. I memorize almost nothing, but I remember the broad strokes of a lot of things. This allows me to be creative.
In a way, memorization is a severe risk: if you memorized something before it changed, for example, your creativity may not mean much.
What I was trying to convey is that fundamentally learning is memorization, whether conscious "rote memorization" or more less-intentional committing to memory from doing an activity
And that recognizing this allows us to speed up the process of learning fundamentals
Which in turn enables creativity as most people see it
Conscious memorization and rote memorization are two different things. I can intentionally put things into memory without doing rote memorization - or even intentionally avoiding rote memorization. The techniques you use also give you different results in terms of whether or how you use the memorized concept or word in foreign language. (For example getting the effect where you can translate a word between foreign and your language, but can not use it foreign language sentence and do not understand it in context without translating.)
You state that "learning is memorization" but I don't think that is true. Of course learning involves something being persisted in one's brain, but stating that memorization and brain persistence persisting are synonyms seems like an incorrect description.
This is like saying that programming competence comes from breathing air.
Creativity (measured by Openness in the Big Five) is a fundamental personality component that cannot be altered, in the same way that your working memory and IQ cant be altered (unsurprisingly they are closely related)
I heard this idea recently: creativity needs shallow knowledge over a vast range of concepts. It’s probably not “just” that, but I can see how the vast range helps.
OP and others here are stretching the definition of “memorize” to mean “anything that leads to something being retained in memory.” I reject this idea.
The trauma of burning your hand on a hot pan creates a memory you won’t soon forget, but almost no one would understand it as an act of memorization.
Memorization to me refers to a set of cargo-culty “learning” practices wherein we believe that by using language to drill exposure to an abstract representation of a concept, that somehow we will absorb the concept itself.
We do this mainly because experts suck at empathizing with learners and fail to understand that the symbol has meaning for them but not for the learner.
It’s the difference between drilling vocabulary flashcards and actually reading, listening, or talking to someone.
Young children do not use vocab flashcards to learn their L1. They aren’t being “drilled” to learn “mama.” They have actual needs in an actual social context and attend to nuanced details of that context to make complex statistical inferences about the world, their perceptions, and their body. Mostly subconsciously.
Yes, there are specific areas where drilling can help us accelerate or catch up. Many kids seem to need explicit phonetics instruction in order to make the leap to reading words. Phonological speech interventions are often drill-like. Practicing musical scales does make you more fluent in improvisation. Drilling the mechanics of a repertoire piece frees your mind to focus on higher-order expression and interpretation. They’re valuable, they have a place.
But this is just a small slice of learning. It’s disproportionately important for passing tests (And getting hired at tech companies!), which to me is the crux of the issue.
If I had to reformulate OP’s argument to something I can get behind, it would be more about deliberate practice or “putting in the reps.” This is also often boring, and differentiates highly successful people from average performers. But it’s a broader and more purposeful set of activities than “memorization” would imply.
> We do this mainly because experts suck at empathizing with learners and fail to understand that the symbol has meaning for them but not for the learner.
I hear you; but teaching deep expertise is really hard. We can use your example of a child learning their first language. They will really understand it. But people are famously, hilariously terrible at teaching their native tongue. We know how to conjugate, and how to use verbs and adverbs and all the rest. But it’s all intuitive - we have no symbolic understanding of it. If that’s the case, we can’t explain it in words.
Here’s a weird fact: if you look around the room you’re in now, I bet you know what it would feel like on your tongue to lick everything you see. We probably learned that in the “put everything in your mouth” baby phase.
You are an expert. But if you wanted to, how would you teach that? I think the learner would just have to go lick a lot of things for themself.
I believe a lot of real learning is actually like that. When I taught programming, I think I was a frustrating teacher. My students would ask things like “what’s the best way to structure this program?” And I would say “I don’t know. Let’s brainstorm a few ways then you should pick at least one and try writing it like that. Figure it out in code.” I think you become great at programming by licking all the programs you can find. Same with music and art and languages (go have conversations with native speakers).
There is only so much the best teacher can teach. Sometimes you just have to walk around licking things.
> There is only so much the best teacher can teach. Sometimes you just have to walk around licking things.
There's being taught and then there's exploring and practice. It's the exploring and practice that is often missing today, IMO because we all want everything to happen so quickly.
In my sport of choice, jiujitsu, there's been a shift happening in teaching. Traditionally the teacher would show a move or two then have the class drill the move. The shift now is more conceptual where the teacher might say the goal is get chest to chest and then have the students operate at 20% figuring it out on their own. While there are some pretty vocal camps on both sides, I think having both methods is the best to learn from. Being taught a solution to a problem a student discovered on their own during discovery is when learning really takes hold IMO.
Finally, the best way to really learn that I have found is to teach something.
> I hear you; but teaching deep expertise is really hard.
No it's not; this is the point of apprenticeship. It's just not compatible with the educational institutions people actually pay for.
> I think the learner would just have to go lick a lot of things for themself.
This is not contradictory. If your idea of teaching is just lecturing at people and occasionally verifying specific knowledge, obviously they're not going to learn very much. Your job as a teacher is to facilitate a growth curve of increasingly more difficult and unwieldy problems to solve.
> Here’s a weird fact: if you look around the room you’re in now, I bet you know what it would feel like on your tongue to lick everything you see. We probably learned that in the “put everything in your mouth” baby phase.
You are an expert. But if you wanted to, how would you teach that? I think the learner would just have to go lick a lot of things for themself.
This is such a weird and amazing analogy. Thank you for sharing it! I don't know if I've ever heard this concept explained in quite this way before.
I quite liked your licking metaphor/symbolism. In fact, if I find out that something tastes good after licking it, I'll go even further. I swallow it. Ha ha.
My intuition is that creativity is a more of a feeling one has, rather than the result of some teaching method or cultural background. It seems very closely related to one's interests and also the novelty effect. Just think about the first time you experienced something, like a great song or your first love. The mind is just bubbling with thoughts. How can that feeling be experienced again with the same object like it was the first time? It can't. It's never the same as the first time. Even if we can mentally forget, the subconscious somehow remembers it's had that experience, so it isn't fresh anymore.
So, yes, you need to make connections between existing things to come up with something new, but once the fascination with a thing dissipates and it becomes familiar, any further creativity in that area quickly dies off. Our interest just moves on.
Perhaps the really creative people are just the ones who have this ability to see familiar things with ever fresh eyes?
Kinesthetic learners know exactly what you are talking about. Others may have no clue. Just a moment ago I identified which pill bottle I needed to open by shaking it. The sound prompted an image of the pill shape and density. This interconnectedness of sign systems and our human ability to read them can truly emerge in brilliant every day ways. I agree nothing replaces experience, especially for kinesthetic learners.
Yes. Tacit knowledge is hard to teach.
All those really learning a foreign language get to the point, where they "chug it" and jump into the deep end, surrounding themselves with native speakers and going to that learning by context mode. Its the toughest phase, but works.
Love this metaphor. To truly learn we must have a lick for ourselves, pun not intended.
The EEs I have known that carried around a card with:
because they couldn't remember it were all bad at EE and bad at math.If you can't remember the pieces making up a concept, how are you going to remember the concept?
> It’s disproportionately important for passing tests (And getting hired at tech companies!)
I don't remember anyone who couldn't pass tests but was really a great engineer.
BTW, one of the tests fighter pilots go through is they are blindfolded, and then have to put their hand on each control the instructor calls out.
I also have some written tests for certifying pilots. There are questions like max takeoff weight, fuel burn rate, max dive speeds, etc. Stuff a pilot had better know or he's a dead pilot.
> There are questions like max takeoff weight, fuel burn rate, max dive speeds, etc. Stuff a pilot had better know or he's a dead pilot.
You don't want a pilot who is creative when it comes to max takeoff weight.
Obviously there are good reasons to memorize certain things, creativity just isn't one of them.
> I don't remember anyone who couldn't pass tests but was really a great engineer.
But there are plenty of people who can pass tests and are terrible engineers. When people talk about memorization they talk about remembering the words without understanding the concepts, such a person is no better off than the person carrying around a text note with the concept written down.
When you understand these concepts you will remember those formulas, but memorizing the formula doesn't make you understand it. Therefore creativity doesn't fundamentally comes from memorization, there is something else there.
> I don't remember anyone who couldn't pass tests but was really a great engineer.
This is just confirmation bias. Either you're in a field where you must have a degree, so everyone who couldn't pass tests simply never became an engineer (eg EE) and whether or not the test is a good measure is irrelevant because it nonetheless gates your sample
OR you're in software engineering and the people who struggled on tests work beside you, but they don't tell you about their past performance on tests because you have a chip on your shoulder and they don't want you to look down on them.
I failed a lot of tests in college and now I'm a great software engineer.
Some times the brain is wired to intuit the concept. That's something that fascinates me. You grasp the idea before any articulate explanation. Somebody shows you a problem and rapidly you start discussing solutions with the other person and even go further.
Most of the time memorization is a key role for creativity, the easier you can jump between ideas the more combinations you can explore (seems like the brain is constrained by some cache bottleneck in a way).
> If you can't remember the pieces making up a concept, how are you going to remember the concept?
This is not mere memorization. The GP's point that not everything that makes you remember this is simply "memorization", and certainly all of those things contribute differently to creativity.
> I also have some written tests for certifying pilots.
Yes, piloting is one of the tasks where you absolutely need lots of random information memorized. You should also not do a lot of "creativity" in it.
Isn't that exactly the result of a focus on memorization instead of understanding (yes yes at some level that also involves memory), though? To a certain type of person memorizing an arbitrary arrangement of 3 symbols is quite difficult but OTOH it's easy to remember that current goes up with voltage and down with resistance.
Zero of the EEs I have known need that (where did yours graduate from?)
The mnemonic to remember V = I * R is as a triangle with V at the top and I, R on the second row, the same way you remember the relation Distance = Speed * Time or Mass = Density * Volume in high school. Or any other triangular multiplicative relation.
This is different memories? the one for the formula and the one for the concept. So you can remember the meaning of the word but don't remember how to write it. So can reproduce the formula from your deep understanding, but it is quicker to check it out instead of apply the first principle every time.
Saddest part is that you only need first one and and then with memorization of most basic algebra you get the others.. Not even need to do it symbolically.
> because they couldn't remember it were all bad at EE and bad at math.
> If you can't remember the pieces making up a concept, how are you going to remember the concept?
You touch upon the different levels of knowing it. Yes, having to carry a card with the formulas on it shows no knowledge. But, if you have to memorize the formulas, your knowledge is still not adequate. You're just regurgitating a formula that you memorized so you can plug in numbers. You don't understand the "why" of Ohm's law. Of course voltage is equal to current times resistance, it's obvious by what these things are! It should be as self-evident as "Of course distance is equal to speed times time!"
Another example: You can have the Lorentz transformation formulas memorized but still not really understand the "why" of Special Relativity.
My high school had majors and one of them was effectively EE; the 14/15 year olds in my major had those formulas down inside of 3 weeks... what EEs did you know that couldn't outperform teenagers?
...I mean, I didn't even go to that great of a college and no one would have made it past the second year of EE without memorizing 10x more formulas than that.
I'm just completely lost on how it's even possible to have an EE degree and needing a card. Signal processing classes required math 100x more difficult than that. I had to know quaternions by my second dsp class.
Memorize means "to retain in and quickly recall from memory". Weather that is by synthetic or natural process is irrelevant. From the point of understanding how memory and recall work, yes burning your hand is an act of memorization.
Sure there is a natural repetitive process that leads to base learning like L1 you mention.
On the other hand no one adds or multiples enough in daily life for natural memory formation. Humans consider the skill vital enough that we have developed methods to memorize them. Same for spelling, especially for infrequently used words.
Flashcards used with Spaced repetition isn't cargo cult, it is a well studied, and pretty good method of inducing memory formation.
I’d say that’s the definition of “remember” rather than “memorize.”
To most people I’d wager “memorize” has a strong connotation for the synthetic version only, with an emphasis on a stripping-out of context.
I recognize that stripping away context can be valuable—-drilling a tennis serve over and over outside the real-time context of a game is extremely helpful.
Flashcards are rarely valuable in the same way. For semantically oriented tasks, an impoverished context is usually not very helpful. Receptive skills like letter and character recognition might be an exception. But even then you’ve got to make the leap to reading at some point.
> no one adds or multiplies enough in daily life for natural memory formation
To the contrary, there’s a fascinating study of children in South America who had very fluent mental math skills for making change because they sold fruit on the side of the road. They couldn’t solve the exact same problem in story problem format in a classroom, though. Synthetic contexts usually don’t transfer well to real life.
By far the best way to learn arithmetic facts is to ‘naturally’ use them in service of solving more interesting or relevant problems. Someone who spends the same amount of time doing nontrivial word problems, pattern-discovery projects, playing a game or solving a puzzle involving embedded arithmetic, or just talking about numbers in a group will come out vastly better prepared both for recalling or figuring out arithmetic solutions per se and for mathematical fluency in general than someone who does narrow practice drills. Arithmetic drills are not only a total motivation killer for most people, but also just suck at aiding retention. Time spent on arithmetic drills in school is somewhere between a waste of time and an actively harmful punishment.
If anyone wants some primary school appropriate word problems, let me recommend the collections by Lenchner, e.g. https://archive.org/details/mathematicalolym0000lenc
Also try Kordemsky’s Moscow Puzzles https://archive.org/details/boris-a.-kordemsky-the-moscow-pu...
> Memorize means "to retain in and quickly recall from memory". Weather that is by synthetic or natural process is irrelevant. From the point of understanding how memory and recall work, yes burning your hand is an act of memorization.
I think in common usage the word "memorize" very strongly implies that it's the lossless storage and retrieval of some highly specified sequence of information. No amount of studying American Civil War history would be referred to as "memorizing the Gettysburg Address" unless you could recite the speech word for word.
> On the other hand no one adds or multiples enough in daily life for natural memory formation. Humans consider the skill vital enough that we have developed methods to memorize them.
Isn't that kind of the point for the other side?
Nobody memorizes the answer to 212+457. There are no flash cards for every possible addition up to infinity.
In my experience, the more people think that math is something to memorize, the worse they are at math.
>OP and others here are stretching the definition of “memorize” to mean “anything that leads to something being retained in memory.”
They "stretch" it to its dictionary definition?
>The trauma of burning your hand on a hot pan creates a memory you won’t soon forget, but almost no one would understand it as an act of memorization
It still is a kind of memorization, just not a voluntary one. And such learning is still is a very important function of mental development and evolutionary fitness, that shouldn't just be shunned "because trauma".
>Memorization to me refers to a set of cargo-culty “learning” practices wherein we believe that by using language to drill exposure to an abstract representation of a concept, that somehow we will absorb the concept itself
Well, I see your cargo cult and I raise you tried-and-true.
"Absorbing the concept itself" might take more effort (including personal, for logistical reasons, not everything can be tailored to the individual learner, who might not even care enough for learning compared to all kinds of diversions, and have zero passion for the subjects, even if a clone with the teaching skills of Feynman with the presentation skills of Tonny Robbins, and the passionate conviction of Jean D' Arc was to present it to them.
But absorbing the concept is not enough, there needs to be instant recall, or at least fast enough recall) of all kinds of facts and factoids and tables, and also this "absorbing" also needs to encompass boring concepts, that are nonetheless crucial, if one is to be succesful in anything technical or scientific, or generally creative in any sort of organized way that combines concepts and information (not just Jackson Pollocking away).
> They "stretch" it to its dictionary definition?
To memorize means to commit to memory. It is an action. But memories are created by not just actions, but experiences. Experiences can create memories without you having committed them; without you having memorized anything.
> They aren’t being “drilled” to learn “mama
They actually are, thou. Small kids are repeated the same things over and over, hundreds of times. They watch the same Bluey episodes dozens of times, get them read the same books dozens of times, listen to the same songs dozens of times.
They drill in their own ways, but they drill.
> It’s the difference between drilling vocabulary flashcards and actually reading, listening, or talking to someone.
Those are not opposite activities. Drilling vocabulary flashcard is the most efficient way to start being able to read/listen/speak; and it's not even clear from research that output (speaking and writing) is useful at all for learning.
Also good luck learning to read Chinese or Japanese without rote learning a few hundred characters. Even native speakers learn them by repetition. You can't be serious advocating "just go read stuff" as a way to learn a language. Some foundation, acquired by explicit learning, is required to even start reading.
> Young children do not use vocab flashcards to learn their L1.
That's a bad argument: the life of a toddler is vastly different than an adult. A young child has basically nothing but figuring out what's going around him (including language) at least 16 hours per day, every day. An adult has much less time for that.
> Drilling vocabulary flashcard is the most efficient way to start being able to read/listen/speak
This isn't actually a settled matter. I did a literature dive a while back and found that drilling vocabulary flashcards shows the highest benefit on artificial recall tasks (like multiple choice tests), and over relatively short time scales (days to weeks). Studies that looked at longer time scales (months to years) and more organic tasks generally showed mixed results. Which I generally interpret as a sign that the literature in question is highly susceptible to the file drawer effect.
And that in turn suggests that the magnitude of flashcarding's value for this kind of thing has a lot to do with your goals. In a nutshell, are you more interested on the science on how to develop communicative proficiency over the long run, or are you more interested on the science on how to get a good grade in class?
I'm studying Chinese right now, and I do use flashcards, but it's not because I believe it's the best way, per se. It's because it's a convenient option for reviewing characters and words that don't appear often in the reading material I have. When they do appear often in my reading material, I find (anecdotally) that it takes a lot fewer organic repetitions to get the character or word to stick in my head than it does with flashcard repetitions.
It's also worth mentioning there's no particular reason to assume Chinese and Japanese schools are any less likely than schools elsewhere in the world to cling to inefficient pedagogical techniques out of a sense of tradition. So one can't necessarily assume that the way they are doing it is the way they ought to be doing it.
> That's a bad argument: the life of a toddler is vastly different than an adult.
Also it takes a surprisingly long time for children to learn language. I used to think it sorta-kinda happened over a year or two, but having children myself revealed how wrong I was.
I have written up a fairly hefty dictionary of words they mispronounce or even just invent on their own because they don't know the one in their native tongue. My oldest rarely contributes individual words to this dictionary anymore, but he still improvises expressions and idioms.
> A young child has basically nothing but figuring out what's going around him
Right—not simply memorizing what's happening around them. Those are fundamentally different activies. That is the gist of the parent comment's point.
They've also acknowledged expressly that rote memorization techniques are "valuable" and "have a place."
> and it's not even clear from research that output (speaking and writing) is useful at all for learning.
Only if your goals don’t include being able to speak or write.
> You can’t be serious advocating “just go read stuff”
True, not “just.” And Chinese is particularly tricky because the ideograms convey little to no phonetic information. Even so, almost any activity I can imagine seems superior to traditional flashcards. Photo flashcards (vs. translation), listening along to highlighted text or closed captions, deciphering street signs or memes, even the written drills you mentioned. (Or better, “write 5 phrases that all start with character X”). Our brains crave meaning, and flashcards offer very little of it.
> even native speakers learn them by repetition
No argument there, most everything is learned by repetition, but I’m interested in context. Native speakers already know the verbal form of most words they’re learning to write, even in Chinese. I’d argue the meaning is stronger.
> life of a toddler is vastly different than an adult
True. The scale of their learning tasks are much bigger than ours. They have to learn that they exist, that their family exists, that they can vocalize, that language is a thing, that they want and need things, and that they can get them by communicating.
> An adult has much less time for that
I think this is a good entry point to the core of the issue for me—-small children don’t “set aside time to learn,” they just learn. You and I do this also, though it’s less novel and flexible and therefore maybe less salient. I think we place too much value on structured or synthetic learning as “real” learning when in fact it’s often extremely inefficient compared to our natural learning tendencies.
There’s a spectrum of structure, starting with what we choose to pay attention to, to an open-ended “study time,” to guided classroom activities, to timed math drills. Flashcards are at the extreme reductivist end of that spectrum. I suspect we like them because they’re easy to understand, uniform, predictable, and convenient to create and use. Creating more effective learning opportunities and supports is substantially harder, but generally worthwhile IMO.
Vocabulary flashcards are not all that efficient way to start being able to read/listen/speak. They teach you translation rather then meaning directly, you don't get context or the "context" is super repetitive sentence and so on.
And plus, general recommendation is to learn words elsewhere and just put them into anki to not forget.
Some people like it, but it is not the only or the most recommended way to learn speak and write.
> Also good luck learning to read Chinese or Japanese without rote learning a few hundred characters. Even native speakers learn them by repetition.
I've tried this for years with Japanese kanji and never really got very far. Just didn't work well, they largely were just a big blob of lines.
Then I found an Android app (Kanji Study) that mixes this in with informational screens that break down kanji into radicals and puts them alongside a bunch of multi-kanji words and uses them in sentences so we can see them in context, and it's actually been working.
You are changing the definition of memory we have had for like 25 centuries in the West. That is your choice but there is no stretching on the other side.
“To me” is not what defines what something is.
> The trauma of burning your hand on a hot pan creates a memory you won’t soon forget, but almost no one would understand it as an act of memorization.
I reject you rejection. You need to provide a reason something learned via pain is fundamentally different from something learned without pain. We used to beat school children for not learning their lesson.
Drilling vocabulary flashcards is very helpful in learning a language. As is reading, listening, or talking to someone. To learn a language well, you need to do both. Although children ultimately reach the level of being a native speaker, they do not do this very quickly considering that they say their first words when they are about 1 year old.
haha, no man, the "creativity" cargo cult should be entirely ignored. You shouldn't even consider what mechanistic formula they've invented. If you do you might buy into it and your creativity will be less your own.
Never listen to the grownups, everything they say is a lie.
Makes total sense. Kinda like how we can't train an LLM how to speak by just giving it a dictionary.
I am not commenting on or agreeing with the OP, but your response is false. LLMs aren't given just a dictionary, and they do not know how to speak. Speech implies grasp of semantics. There is zero semantics in a block of text, only, according to some interpretation, various textual correlations.
> It’s the difference between drilling vocabulary flashcards and actually reading, listening, or talking to someone.
You need the 'flashcards' before you can read, listen or talk. Go try reading a book where you don't know most of the words. Heck you need 'flashcards' before you needs 'flashcards for words'. You need to memorize the alphabet first. Try reading a text where you haven't learned the writing system.
> Young children do not use vocab flashcards to learn their L1.
Because they can't read.
> They aren’t being “drilled” to learn “mama.”
Obviously you aren't a parent. You think a child magically decides one day to say mama? Or do you think it's the mother constantly saying 'mama' to the child until the child 'remembers it' and repeats it?
> They have actual needs in an actual social context and attend to nuanced details of that context to make complex statistical inferences about the world, their perceptions, and their body.
What? Complex statistical inferences about the world?
> Heck you need 'flashcards' before you needs 'flashcards for words'. You need to memorize the alphabet first.
I've a toddler who can read 3 paragraphs of 3 sentences each, and then tell you the details of the story he read[1]. He is 4y6m, right now. He has never learned the alphabet or the names of the letters (A,B, C, D, etc). He has only learned the sounds a letter or sequence of letters make for certain patterns.
You most definitely do not need to memorise the alphabet in order to learn to read!
Teaching children the alphabet before teaching them reading makes it a lot harder for them to learn actual reading.
[1] I've seen kids as old as 7 get confused by a book with no pictures, and he sails right on through because I taught him to read (using the DISTAR alphabet), and made sure none of our daily lessons had even a single picture in it.
You should be aware that people are able to become fluent without ever using flashcards.
Kids say mama almost universally and regardless of their local language because it's an easy sound to make.
How are you defining a "flashcard"?
> ...experts suck at empathizing with learners...
Or maybe we just don't want to coddle them. When has learning anything been easy, and why do you expect people to be able to acquire new knowledge and skills without putting in the effort? It shouldn't be grueling, not for its own sake, but yeah, you might have to stare at a compiler error for a few hours or even a few days before you figure out what's broken. Truly, how else are you supposed to learn if you don't, eventually, do it yourself?
I'm so sick of this anti-expert, anti-knowledge attitude. It's why we have bootcamp juniors being thrown into otherwise-senior roles, with laughably predictable consequences for the field.
Perhaps I worded it too emotionally. I mean that experts struggle to remember what it was like before they understood something. It’s very common for experts to ask novices to make leaps that they aren’t capable of making, because they seem natural or obvious from an expert POV.
I’m all for hard work; learning is usually very hard work. I also think we need expert guidance.
But let’s make the difficulty useful/effective rather than counterproductive.
This attitude seems unrelated to the topic at hand quite frankly. Experts suck at empathizing with learners not because of this spite, but often because it's actually quite difficult to switch gears in language and understanding. It's a completely different way of sharing knowledge, where you have to explicitly express things that are just assumed shared understanding among colleagues.
Also, to answer your questions in a very simple way: the entire reason you even became an expert is because another expert somewhere along the way gave you an easy in to the knowledge, they coddled you. This is why you can "stare at a compiler error for a few hours before figuring out what's broken". Without that expert, you wouldn't even understand what a compiler even is.
Parent is saying there's a big difference between being an expert in a field and being a good teacher.
"a flash of inspiration connecting internalized concepts"
Well, okay, but rote memorization is neither necessary nor sufficient to internalize concepts.
One of the reasons people make fun of the author's approach to creativity is that systematic memorization fundamentally can't teach taste—so the systematic approach reeks of awkward, try-hard, low-brow, tasteless art.
More broadly, memorization doesn't help much with any sort of tacit knowledge, not just taste. I just figure taste is especially important in creative endeavors. That's definitely the case for programming! Memorization in programming gives us architecture astronauts and design-pattern soup rather than elegant code.
For what it's worth, I do think that it is useful and important to have a good mental model of what expertise is and how you can develop it. Memorization might be a component of this, but it's going to be a small component at most. I expect that realistic practice with fast feedback and expert mentorship matters far more. (If you're curious, I found the book Sources of Power by Gary Klein gave me a good way to think about how expertise works.)
At the same time, memorization has a real cost: it takes time and it's frightfully dull. For me, at least, trying to memorize something without context is not just ineffective but also totally kills any intrinsic motivation I have for whatever I'm learning. Sometimes a bit of memorization is unavoidable, but I've found that to be relatively rare. Otherwise, my time is generally better spent on some sort of practice in context.
Thanks for reading and the response!
One of the points I'm trying to make is that taste and elegance fundamentally stem from an internalized heuristic -- which at it's core is memorization.
I understand the connotation of "memorization" evokes an image of blindly memorizing without connecting, but isn't the tastefully developed expertise just memorization of a better heuristic?
I don't think I can agree, as an extremely creative person with extremely bad memory - to a point where I pretty much never memorize anything, whether intentionally or by accident.
What I find instead, is that by just processing novel information, especially if I focus on analysing it, my brain internalizes insights and builds model of that type of thing, allowing me to either imperfectly reconstruct what I've seen, or to come up with an infinite array of permutations, extrapolations, etc which is where the real ideas come from.
Further, ideas crucially revolve not around just the information itself, but the "feel" for what role they play in the whole, how well they do it, in what way they're notable, etc.
In fact I'd straight up claim that memorization is antithetical to creativity - a perfect ML autoencoder or GAN would just regurgitate the training data. Creativity comes from generalisation while memorisation is analogous to overfitting.
That's true if you broaden the definition of "memorization" to cover all learning, but "learning is necessary for creativity" would not be a particularly interesting thesis.
Expertise is the result of learning from past experience, both in developing an internal intuition for what you're doing and in having past patterns to draw upon. To the extent that experts have simple easily verbalizable heuristics, these are largely post-hoc attempts at explaining their intuition rather than an accurate reflection of how they make decisions.
And, in fact, experts can't even always do that: it is perfectly possible for experts to make good decisions without being consciously aware of why they are making them, and explaining how to make good decisions is a separate skill from being able to make them in the first place. The book I mentioned has a memorable story about a firefighter who thought he had precognition after pulling his team out of a dangerous situation without any specific indicator of the danger, but I figure a more common example is experts saying they did something because it was the "obvious" or "clean" or "better" way to do it and getting a bit flustered when pushed further.
We can see this in action pretty clearly if we look at advice for, say, writing. There is a lot of advice from good writers but just memorizing and blindly following this advice is actively counterproductive. Advice you can memorize fundamentally must lack nuance and context. We can see this clearly because so many different pieces of writing advice contradict each other and because good writers do not follow any of those suggestions with any consistency.
The same definitely applies to programming, which is why we have both "don't repeat yourself" and "you ain't going to need it", and why new programmers trying to apply either rule (or both!) to a codebase inevitably create a mess. What I've found with programming advice is that most suggestions are either actively wrong or too vague to be useful. (By the time you've learned enough about programming to be able to follow the vague advice, you don't need it very much!)
Why are you attached to the word "memorization" here? Certainly taste comes from experience and learning. Maybe you could argue that all learning is an oblique and imperfect form of memorization—but why argue that at all?
The only reason I can see is if you think memorization could be a shortcut to good taste, which it can't. Acquiring good taste requires broad experience—more information than you can possibly remember—such that you retain a suite of sophisticated intuitions. Cutting that information down to something that can be memorized would require you to (1) already have the intuitions you're seeking to acquire, and (2) be able to express them all in plain English, which, as far as I know, cannot be done. No painter has ever expressed their aesthetic in such a way that a student could memorize that expression and then have the same creative sensibilities as the original painter.
Ultimately, there's no substitute for the process of simply consuming lots of art while paying close attention to what you like about it.
>One of the points I'm trying to make is that taste and elegance fundamentally stem from an internalized heuristic -- which at it's core is memorization.
seems to me there is a relatively big inductive gap there, you believe that there is an internalized heuristic and at its core is memorization, you may even have some evidence that this internalized heuristic has strongly informed your development of taste, but it is pretty difficult to make an argument that is the case for all people.
Aside from that I would say that "internalized heuristic with memorization as the core" puts everything on nurture and no input of nature - which I am pretty much in the camp of combinations of nature and nurture creating the person - of which taste must surely be a big component.
Schmidhuber reached your conclusions first.
Driven by Compression Progress: A Simple Principle Explains Essential Aspects of Subjective Beauty, Novelty, Surprise, Interestingness, Attention, Curiosity, Creativity, Art, Science, Music, Jokes
https://arxiv.org/abs/0812.4360
not the parent poster but I think I agree with your perspective here. The alternative is that some individuals' taste or sense of aesthetics is somehow innate and unmoored from the statistics of the things they experience. There may be something to this, but for most practical purposes I would agree with your point.
also a question - if you have better long term or short term memory how does that affect taste? How does it affect creativity, if all of these things are essentially memorization you would have to assume that people were more creative and had better taste the greater their ability to memorize things, which in the case of taste especially seems slightly absurd.
In the case of creativity it may be easier to make an argument - but surely you can find people who seem more creative with less ability to memorize.
> Memorization in programming gives us architecture astronauts and design-pattern soup rather than elegant code.
Elegance is probably orthogonal to creativity, and likely follows from some kind of minimization principle, like minimum program length. You are effectively distilling the "essence" of something from all of the noise.
Creativity seems different, more like novelty, and creativity following some kind of remix of memorized elements + some randomization seems very plausible.
You can create something novel but not elegant, and something elegant but not novel, and you can distill an elegant version of something novel that your or someone else created and that's the best of all creations.
I completely disagree with your assertion that "...rote memorization is neither necessary nor sufficient to internalize concepts."
I would recommend reading the book Moonwalking with Einstein. There is a lot of discussion there on how memory is linked directly to creativity, and to understanding concepts deeply.
---
A choice passage:
"...If the essence of creativity is linking disparate facts and ideas, then the more facility you have making associations, and the more facts and ideas you have at your disposal, the better you'll be at coming up with new ideas. As Buzan likes to point out, Mnemosyne, the goddess of memory, was the mother of the Muses.
The notion that memory and creativity are two sides of the same coin sounds counterintuitive. Remembering and creativity seem like opposite, not complementary, processes. But the idea that they are one and the same is actually quite old, and was once even taken for granted. The Latin root 'inventio' is the basis of two words in our modern English vocabulary: inventory and invention. And to a mind trained in the art of memory, those two ideas were closely linked. Invention was a product of inventorying. Where do new ideas come from if not some alchemical blending of old ideas? In order to invent, one first needed a proper inventory, a bank of existing ideas to draw on. Not just an inventory, but an indexed inventory. One needed a way of finding just the right piece of information at just the right moment.
This is what the art of memory was ultimately most useful for. It was not merely a tool for recording but also a tool of invention and composition. "The realization that composing depended on a wellfurnished and securely available memory formed the basis of rhetorical education in antiquity," writes Mary Carruthers. Brains were as organized as modern filing cabinets, with important facts, quotations, and ideas stuffed into neat mnemonic cubbyholes, where they would never go missing, and where they could be recombined and strung together on the fly. The goal of training one's memory was to develop the capacity to leap from topic to topic and make new connections between old ideas. "As an art, memory was most importantly associated in the Middles Ages with composition, not simply with retention," argues Carruthers. "Those who practiced the crafts of memory used them---as all crafts are used---to make new things: prayers, meditations, sermons, pictures, hymns, stories, and poems." ..."
Great book, motivated me to then read The Art of Memory by Frances Yates.
Although I'd say traditional mnemonic devices like memory palaces are basically linear information storage and recall devices. This can create issues in building a flexible web of information, because loci or the order of the path can become dependencies and you can run out of unique spots in a given space, leading to memory interference.
Even spaced repetition methods (e.g. Anki) tend towards fragmentation of micro-ideas. Its perfect for terms, languages, and simple one question -> one answer ideas.
I've found a hybrid method of images, nested loci and spaced repetition to be most useful, because its flexible over time, and preserves relationships of ideas.
(Context: I co-founded a SaaS in this space: www.sticky.study)
Great passage -- this is exactly what I was trying to get at, though they've described it with much more eloquence and historical backing.
Have never heard of this book but adding to my list now!
Didn't Einstein say that don't memorise what you can look up? E.g. nothing nowadays since we have the Internet.
The book Make it Stick taught me that this Don’t Cramp My Style With Your Boring Rote Learning (Man) attitude is prevalent in teaching. At least American. They argue that it is wrong for the same reason that the author does.
But saying this to a programming crowd must be the most futile thing. At least instrumentalists have to rote train their muscle memory. That lowest bar has to be passed, even if it’s just three chords.
But the article isn’t about programming creativity though. It is a general concept. But if honing in on the mythical lone-genius activity (geniuses never practice in a structured way) helps you win an argument then so be it.
I think it is more of becoming fluent with primitives that can be composed in versatile ways. I can see how that can be poorly understood as memorization.
The main implication is that if what you are “memorizing” is not easily composable, then you won’t be able to apply them broadly or creatively.
However, I disagree with the author on what creativity is, although his definition is one experience of a creative inspiration.
Interesting perspective. I do agree that there are people out there who develop a distinct "taste", but I can't tell if this refers to a "style", an emergent property of multiple "habits", etc? I've always wondered how one develops their "taste".
Also, would you consider a subconscious habit "memory"? What's the difference between the two?
Counterargument: the alphabet.
> One of the reasons people make fun of the author's approach to creativity is that systematic memorization fundamentally can't teach taste—so the systematic approach reeks of awkward, try-hard, low-brow, tasteless art.
Well... can you think of an artist who didn't have a deep knowledge of their art-form before they pushed it forward? Three that jump out for me, in no particular order, are Picasso, Borges and Jack White. After all, great artists steal.
There's a big difference between "artist who didn't have a deep knowledge of their art-form" and "artist who didn't follow an explicit system to memorize a bunch of rules to make their art".
Wait, why do you think Picasso didn't have deep knowledge? He studied both at the School of Fine Arts in Barcelona & the Royal Academy of San Fernando in Madrid, for ~5 years before moving to Paris.
Borges was incredibly talented, but it's worth keeping in mind his dad was a writer too.
Good art very much relies on being exposed to lots of other good art first. I don't know that rote memorization is the best way to achieve that, but you definitely need that exposure.
Picasso was a hack. People often cite his "early masterpieces", but those pieces are pretty mid in the context of 19th c. painting.
I think that argument is that these artists did not memorized rules or previous pictures and then applied them. They did put a lot of effort into learning, but that is different claim. If you define "memorization" as "any learning of anything", then the word is kind of useless.
Pollock is often regarded as pushing painting forward, for example
> Well, okay, but rote memorization is neither necessary nor sufficient to internalize concepts.
Of course it is. It's how every human child learns initially. By rote memorization. How does a toddler learn how to say mama? By constantly hearing and repeating it. How does a kid learn their ABCs? Rote memorization is the basis of all memory.
> Memorization in programming gives us architecture astronauts and design-pattern soup rather than elegant code.
Dumbest thing I've ever read. You write programs well by doing and remembering. Same with writing. Memorization is the necessary component to programming well. In other words, you program well by remembering elegant code.
> For me, at least, trying to memorize something without context
After the basics, most memorization is contextual.
> At the same time, memorization has a real cost: it takes time and it's frightfully dull.
Oh dear. Something isn't fun all the time. What a childish worldview. It's more fun to eat candy and drink soda than eating 'dull'. It's more fun to sit and watch youtube than to workout.
> Sometimes a bit of memorization is unavoidable, but I've found that to be relatively rare.
Relatively rare? In order to be competent in anything, you have to memorize lots. You can't write a good essay without having memorized much of the material. Trying reading a book where you have to constantly look up definitions of words because you lack the vocabulary. Try having a conversation with someone who has to constantly look up words because he lacks the vocabulary. Try having code review with someone who doesn't remember anything about their code.
> Otherwise, my time is generally better spent on some sort of practice in context.
Why? Because it helps you remember?
To the idiot ( probably OP ) who downvoted, try coding without having 'memorized' the keyboard. The anti-intellectual, anti-hard work, anti-memorization agenda pushed by some 'people' online bears looking into.
Strong points, but insults and emotion aren't how we do it on HN.
The most striking comment was this:
> Growing up with Indian parents in California, I was exposed to both. My mom would write daily Kumon sheets out by hand for me to do, and teach me from Indian textbooks from the same grade (which were much more advanced than the US equivalents). The result was me breezing through the US school system without much thought.
Ukrainian refugees I know are finding the same things in the UK school system, where the maths is much less advanced. Philippines schools, meanwhile, have better discipline and more motivated students.
I conclude that Western public education is in a bad state, and this is a source of chronic social weakness.
With the caveat that I haven't looked in a while, the rot seemed very anglo-specific. Coming over from Germany almost three decades ago, it was amazing how much US text books just didn't cover.
There were a few developments in Germany that pointed in the same direction, but there a large gap at the time. Meanwhile, even back then, eastern Europe certainly had even higher standards. (I replaced a lot of math textbooks with a copy of Bronshtein & Semendyayev)
The influence of religious and conservative types has weakened US textbooks. Do a deep dive into how Texas can dictate what is presented in schoolbooks nationwide.
> Ukrainian refugees I know are finding the same things in the UK school system, where the maths is much less advanced.
This doesn't appear to be reflected in PISA scores (489 UK, 441 Ukraine)
Maybe because UK education system knows about this PISA and Ukraine and other countries don't even care?
For example one thing that puzzles me in western education is this reading comprehension. What exactly is this for? I studied in Russian school/university and we had nothing like that.
Reading means you should understand what you are reading, reading without comprehension is just nonsense for me.
Also, this problem is a vicious cycle. Bad math students become bad math teachers. I believe that one has to be significantly good at a subject in order to teach the basics well. In school I got a feeling that many math teachers did not know a chapter of math beyond what was in the textbook.
Doesn't help that in the American economy, good mathematicians can earn 10x more than a math teacher's salary.
Why do you assume Ukrainian math is memorization based? They actually do a lot of problem solving, they are not doing rote memorization.
Indian math system is not memorisation based either.
Math anywhere in the world is all about how many patterns you know/seen. And inventing new math is all about changing one small thing about the stuff you already know and see if it remains logically consistent.
Knowing what to change largely depends on the stuff you already know. To that end even if you don't memorize facts, you still need to memorize patterns.
You have to memorize something. Nobody invents the Apple pie from scratch.
I was the sort of person who did not believe in memorisation as a solution for anything. Then I tried getting really good at spaced repetition for a year (yes, it is a skill that needs to be trained for good results) and I've completely changed my mind.
Spaced repetition allows me to become proficient even in things I don't get the natural opportunity to practise daily, so that when the day comes and I need them, I have some level of knowledge already. This has happened to Kubernetes troubleshooting, statistics, PowerShell windows programming, and traffic engineering just in recent history.
I have yet to publish some of these, but I have examples from statistics:
https://two-wrongs.com/intuition-and-spaced-repetition.html
https://two-wrongs.com/inventing-fishers-exact-test.html
The latter is certainly creative in my book, although it does imply creativity within strict bounds.
I'm similar. This from your top link stood out to me:
"It’s a little like building with lego bricks or something – spaced repetition helps ensure all the tiny pieces are in the right place, so that the big castle can happen without structural integrity issues."
The book Make it Stick (by Henry L. Roediger III) had a similar idea they called 'Structure Building'. Very similar to what you described, more experienced and effective learners were creating mental schemas of how the little, but crucial parts of a subject fit together, and successfully cut through the noise.
Structure Building was associated with interleaved practice (shuffling of problem types) and spaced retrieval practice.
I really enjoyed both blog posts, thank you for sharing! And I have to say, your explanation of the subexponential distribution property was remarkably clear for someone without a background in statistics :)
Would you mind sharing the flashcards you generated to build this intuition? I've been using Anki for a while and really trying to focus now on improving my prompt writing; would love to see how you managed it for this problem.
As much as I would like to, I think getting to that understanding required at least 500 flashcards on general statistical and probability concepts, ranging from fundamentals to extreme value theory. Most of those are only barely relevant at face value, but still contribute to understanding.
It's not that I set out to understand this specific thing but that I had studied statistics with flashcard support for a year and that happened to work after a few attempts.
You're always memorizing something at some level, even in math where you can derive so much after memorizing some core concepts and deductions.
This is similar to a loose life-thought i've had for a while, though i lack a catchy phrase for it lol.
My thought is: You're always practicing a behavior whether you like it or not.
Your mind is always setting you up to do more of whatever it is you're doing now. Both physical and mental. To do it more efficiently. With more ease. With more frequency. etc.
It's good motivation for me to mitigate a lot of negative behaviors. Angry in traffic, self anger, etc. If i can reason at least, of course. As i'm not interested in doing a lot of things more than i am, so i should avoid doing them now - if possible.
Math proof and derivations are a bit like remembering a walking route. You've seen the start and end, and the main turns taken, and there's also a general "walking" skill you need.
My argument is that it is worth memorising also the derivations, rather than re-deriving from scratch each time.
Meorising the derivation makes it easier to derive a second-order derivation, and so on. At some level of abstraction, going from first principles becomes prohibitively expensive and caching intermediary results, or so to speak, unlocks that again.
Anki flash cards?
I use org-drill in Emacs but it's the same idea, yes.
The trick is not so much which software or settings one uses, but writing high-quality prompts.
I think memorization gets a bad rep because you need to be acutely aware of what you're memorizing, like memorizing the sequence of an answer sheet instead of core concepts. But when done sufficiently rigorously, the foundations of memorization make room for higher-level critical thinking and reasoning.
Practice is an oft suggested solution to developing mastery, but I did like how the article framed it: creating subconscious heuristics and memory.
Absolutely, especially in real world application. If you don't have the ability to pull on fundamental ideas anywhere, anytime, then have you really mastered the learning material?
Right. I noticed this acutely in an abstract algebra course. We learned several different proof methods, then the exam was just "prove these theorems with the tools you have". I'd never been challenged with math like that before. I mean, I bombed it lol, but nobody was going to pass if they didn't remember, say, how to do a proof by induction or what it means. At some point, you need to be able to recall this information. Maybe the psychologists categorize these things differently, but I'd argue it's clear that some form of memorization is necessary for the task.
Couldn't have said it better, exactly -- the negative connotations of the word prevent us from recognizing what powers learning at its core
But imo acknowledging this unlocks greater speeds and gets us to the "fun part" quicker
Is there anything substantive here?
It’s just a bunch of arbitrary unprovable assertions.
Everyone here seems to have, broadly speaking; neither a) the qualifications to knowledgeably comment of the (honestly poorly understood, afaik) function of “creativity” or b) anything more meaningful than “here is my naive personal lived experience and opinion” to contribute on the topic.
It’s just armchair psychology.
If you want to wax philosophical, by all means, but I think anyone taking “thoughtful insight” away from this article or thread is fooling themselves.
Whatever the epistemic quality of the article is, it's triggered some interesting discussion here which I think is valuable. No need to denigrate talking about human experience with other humans, I think?
You'll experience the creativity outlined in the article directly when you start doing deliberate memorization, i.e spaced repetition. No qualification needed.
I think semi-obvious would be a better criticism than unprovable.
You can't make connections unless you have things to connect to.
You can't recognize (your own discovered/inspired) novelty unless you have memorized normality.
If you are creative in the absence of knowledge of what already exists then that's considered as reinventing the wheel, and not very useful, even it it's Ramanujan reinventing much of established mathematics.
One might talk about it from the perspective of birdsong, which is used by mates to judge sexual fitness. First a tutee bird learns from a tutor bird, and then eventually applies variability to the original song.
It's strongly suspected that anterior forebrain pathway (AFP) may be a source of behavioral variability. We naturally age over time, including our vocal musculature, so in some sense we must constantly relearn how to use our muscles to deliver a song.
When a bird is deafened its birdsong will naturally drift, but when we precisely damage the AFP along with deafening we find that birdsong remains stable for a longer period of time, until of course inevitably it must drift due to aging vocal musculature.
It's pseudoscience, author would benefit from reading (and memorizing) the current scientific literature on learning and cognition.
There's a bad trend, Hacker News gets this kind of blog-style self-promotion every month that gets much comment attention, but the essays are not well-researched and with made-up assertions written by programmers talking out of their lane ("Engineer's disease") and not having done homework on the subject.
> It's pseudoscience, author would benefit from reading (and memorizing) the current scientific literature on learning and cognition.
How so? Can you share what you have read and that is relevant and applicable in real life?
> There's a bad trend, Hacker News gets this kind of blog-style self-promotion every month that gets much comment attention, but the essays are not well-researched and with made-up assertions written by programmers talking out of their lane ("Engineer's disease") and not having done homework on the subject.
You are on the wrong site, this is not a scientific journal there is no need for scientific rigor in every post and comment.
P.S. The real life world is full of events and things happening if you cannot learn by yourself (create theories models on how the world works aka. pseudoscience apparently) and need academic verification for everything. Then you some kind of disorder :/ good luck
> Is there anything substantive here? There is plenty.
If you did not find anything interesting or anything that made you think, in the comments or the post itself. Then is your own failure.
[dead]
Alphazero was very creative, yet it didn't memorize a single move, it just self played. Deep blue was not creative at all, but it was the chess engine that memorized the most moves, todays chess engines are much better at chess and they don't memorize many moves at all, if you dig through their internals you wont find a lot of board states there.
So no, creativity doesn't fundamentally comes from memorization, memorization is neither sufficient to become creative nor is it a requirement. You don't memorize concepts you build models around the concepts. You wont be able to reproduce the exact descriptions of concepts but you will be able to produce something similar that means basically the same thing.
What happens when you forget the models?
All very nice and handwavey, but then you see the user’s current venture is scammy deepfakes as a service, which is about as creatively bankrupt as it gets.
Shame about the national stereotypes as well. There is plenty of creativity in Asian countries. Just bizarre assertions all around.
Historically eastern Asian cultures have placed duty to a whole host of things before oneself, and in many cases the old aphorism "the nail that sticks up gets hammered down" applies as well. Plus Japan and Korea have well established cultures of adherence to tradition and mastering simple, time tested things to a ridiculous degree rather than trying to innovate.
Not true of everyone, but if you compare a culture that values conformity and tradition to a country that values the freedom for the individual and trying new things, of course it's not going to measure up by western standards of creativity.
Have you seen their art and entertainment? I assure you there is whole lot of creativity in there. And it has a whole lot MORE variety then western tend to have.
"Repetitio est mater studiorum" - repetition is the mother of learning.
My creative writing professor, of all people, used to repeat this three times before every class. He was my favorite teacher at any level.
Ah, I knew that phrase but never connected it to the title "mother of learning" (web novel). Thanks for that tiny epiphany
I fear not the man who has practiced 10,000 kicks once, but I fear the man who has practiced one kick 10,000 times.
Bruce Lee
My Arabic teacher liked to say "التكرار يعلّم الحمار" which rhymes and says "Repetition teaches the donkey"
Not the most flattering of proverbs, but it stuck with me.
I genuinely thought creativity was something else until LLMs hit escape velocity and humbled me hard.
After that I realized that creativity wasn't some magical quality that would be hard to reproduce mechanically.
And that also made me a little sad.
But LLMs to date can't really differentiate well between a creative insightful answer, and a nonsensical one. The selection process is still done by a human.
This is true. I hadn't thought about that aspect.
I'm glad you said this -- I felt the same way after making this discovery through my method outlined in this post
It similarly took the magic out of creativity and learning a bit, and made it all seem like work
The main way I've found around it is the joy in being creative once basic autonomy is achieved in new skills
Consciously discovering the heuristic is another fun part
you guys might be interested in the latest Machine Learning Street Talk podcast [1] which right from the start is all about how LLMs are great for creativity, as in novel combinations of trained data (from memorization) -- but are not capable of the reasoning skill needed to verify if any idea is actually plausible given a set of constraints.
[1] Do you think that ChatGPT can reason? https://www.youtube.com/watch?v=y1WnHpedi2A
I have to say, you were naive to ever think that. Rolling a die can produce creativity — and I’m not being facetious.
Creativity, at least in fields traditionally considered to be ‘creative’ like music and visual art, is 90% randomness and 10% retroactively attempting to ascribe meaning to the randomness. And this is coming from someone who hugely values and defines oneself via an ability to be creative.
Also, LLMs haven’t hit ‘escape velocity’; that terminology has spread like a meme (because it’s an attractively dramatic yet easy to understand idea) but isn’t backed up by any science. Maybe one day we’ll have the fabled ‘AGI’, but ChatGPT isn’t it.
This seems to resonate with my experience, although I feel myself bristling due to the baggage of the word memorization.
Although sometimes “memorization” doesn’t happen because you sit down to do it but rather that you keep using the same things over and over when solving problems that they become internalized. I find that to be a more fruitful path towards understanding that I don’t want to call memorization but it is.
Maybe you should first try to separate the negative feelings you have towards the word "memorization" and the word itself "memorization". There's nothing bad about memorization. This sort of negative bias about inconsequential things is something that can easily hold us back from things that could help us further ourselves.
Tell me about it. Trying to get better every day.
Thanks for reading!
And agreed -- it's this exact realization that led me to both this method and title
Imo this negative connotation has made many people refrain from calling internalization what it is
But acknowledging that it's memorization has actually made me more efficient at learning, since I can now consciously look for the heuristic, codify it, and try to commit it to memory
Kind of shocking to see so much angst against memorization here.
Memory has been long thought to be a critical component of intelligence, with elaborate mnemonics systems developed by people to help memorize more things (see Francis Yates' The Art of Memory, and to a lesser extent her book on Giordano Bruno).
I would contend that memorizing concepts is a first step in understanding them. Also, that generally understanding concepts isn't a one and done thing, usually there's layers. Personally I found that memorizing things in math helped me immensely when years later I needed to actually understand the things I had memorized.
A fellow CS student didn't understand a theorem, because he didn't understand any of the three definitions used to state it. We went through the definitions together and suddenly the theorem was "trivial"
That is true understanding. He won't need to remember the theorem, because in his mind, it automatically follows from the raw data.
> The inspirational lightning bolt writers and artists experience can't happen unless they know how to write or draw.
It absolutely can happen (from my experience), but you might not be able to do anything about it. Like I have new melodies and songs pop into my head fairly often, but I'm not good enough at making music to translate those ideas into finished songs (I have made some songs with FL Studio in the past, so it's not totally impossible, but my focus has been on other things).
I also have ideas for stories fairly often too, and while I could write them, they tend to get backburnered for my game ideas, which I also have.
For me, I have plenty of creative ideas, I just don't have the energy or system built up to help me get those out there in a fast enough manner, and I likely won't ever have that, it would pretty much require other people to handle almost all the rest of the work beyond the inspiration and testing out a few things.
As an example, I've made over 60 prototypes for board game ideas in the past five years alone, but still have only gotten one of those games picked up by a publisher. If I were Reiner Knizia, all 60 of these games probably would have been released, because he's built up an engine around him (people willing to playtest all of his designs enough to be polished, and has enough of a reputation that finding publishers willing to publish most of his designs is super easy), and never has to worry about the look and feel or the manufacturing or even the theme of the game, that will all be handled by the publishers.
The spontaneous melody inspiration is probably because you're exposed to music a lot and could easily recompose using those scales, instrument concepts, etc.
And 60 prototypes sounds a lot like practice to me. I wonder if that person you named (who I, a layman have never heard of) made a few games before they got to where they are. All that practice builds up ammunition for creativity.
The consensus I'm seeing reading these comments is that the article is a lot more useful if you replace "memorization" with "deliberate practice". The first implies little deep understanding but lots of practice. The second implies both.
I think if you've ever been exposed to music your mind could come up with different melodies, and who hasn't been exposed to some sort of music in their life. Even if they were the only human on the planet, they'd probably still encounter songbirds, at least.
I disagree that you have to be exposed to a lot of music to even have a chance at coming up with something. Maybe coming up with something good, sure, but not having that spark of inspiration. People can have flash inspiration for garbage ideas too, it happens all the time in startups :P
That being said, the spark might happen more often the more you practice doing something about it, to your point.
Reiner Knizia is the most prolific board game designer of all time. He has 742 games credited to him on BoardGameGeek (some of those are rethemes of older games of his, but he still has probably at least 500 different games released).
About half of my prototypes are about on the level of cleverness and completeness as the lower rated half of the games he has published, at least, and maybe even a couple might slip into the top half somewhere (at his most brilliant his games are way, way better than mine though). But his name carries a lot of weight so he doesn't struggle to find publishers for a good chunk of the games he comes up with.
Meanwhile I've mostly switched back to video game development because at least that way I can easily publish the game myself when it's ready (game development is super hard in other ways, though). I tried for six years and got one signed board game that still hasn't been released after four years. Meanwhile there was a time where I was releasing 1-2 Flash games a year. If I had just made video game versions of my game prototypes instead I probably could have some basic playable version of it out on a website somewhere.
Although I do still submit some board game designs to unpublished game design competitions (I was a finalist in one of the biggest ones, for a design which still hasn't found a publisher) and go to conventions and set up meetings with publishers when I can afford to do so (I couldn't afford to this year, my spouse took a break from work and our income cut almost in half).
Umberto Eco had already discussed this extensively in his paper “the combinatorics of creativity”.
Unfortunately, I couldn't find the English version, but it should be very easy to translate, it's only 16 pages.
http://www.umbertoeco.it/CV/Combinatoria%20della%20creativit...
Thanks for the link. Didn't know the text.
I found an English translation here:
- https://w4nderlu.st/media/pages/publications/combinatorics-o...
Translated by Piero Molino at https://w4nderlu.st/publications
Oh wonderful. I hope you can find time to read it! Umberto Eco was one of the most important modern Italian philosophers.
I love his pragmatic and rational approach.
A Columbia b-school professor has written a lot about this and developed a very compelling framework based on this. Bill Dugan - https://www.goodreads.com/author/list/7325584.William_Duggan.
The first two books are fantastic.
What's more is that memories are just a replaying of neuron connections activating in the brain - and when we are prompted by the world around us those connections will fire in response to the stimulus. Quite similar to how AI neural networks function - which is why I believe that AI can indeed be creative and create "new" ideas
I actually made a video diving deeper into this and comparing responses from people and ChatGPT for a creative thinking problem - https://youtu.be/l-9EUBbktqw
I think your hypothesis here (and probably the entire article as well) is strongly challenged by the 'progenitor argument.' Take humans at the dawn of humanity. Language did not even exist beyond what may have been crude sounds or gesturing and collective knowledge did not fall that far beyond 'poke him with the pointy side.' Somehow we went from that to putting a man on the Moon in what was essentially the blink of an eye.
Training an LLM on the entirety of knowledge at this dawn of humanity and, even if you give it literally infinite training time, it's never going to go anywhere. It's going to just continue making relatively simple recombinations of its training set until somebody gives it a new training set to remix. This remix-only nature is no different with modern knowledge, but simply extremely obfuscated because there's such a massive base of information, and nobody is aware of anything more than a minuscule fraction of it all.
---
As for the 'secret' of LLMs, I think it's largely that most language is extremely redundant. One thought or point naturally flows.... why do I complete the rest of this statement? You already know exactly what I'm going to say, right? And from that statement the rest of my argument will also mostly write itself. Yet we do write out the rest, which is kind of weird if you think about it. Anyhow the point is that by looking at language 'flow correlations' over huge samples, LLMs can reconstruct and remix arbitrarily long dialogue from even the shortest of initial inputs. And it usually sounds at least reasonable, except when it doesn't and we call it a hallucination, but it's quite a misnomer because the entire process is a hallucination.
I agree with the author, at least in my own creative experiences. However, it's more likely the case that 'creativity' is arrived at differently for everyone. I find memorization to be a comforting foundational activity that builds knowledge & confidence, which I can later express creatively.
Exactly -- memorization provides the base for creativity to take place upon
But that creativity can come from many places and in many forms!
There is deliberate practice for skill-building. There is exploratory "making" that fuels originality. There is inspiration hunting and incremental tweaking to get to creative mutation. There is high productivity that triggers eventual ingenuity. I find the article hyperbolic in its thesis and execution especially when it comes to the final hand-wavy bit about how there is more per-capita creativity in non-rote learning.
While its hard to prove or disprove without a long study to prove or disprove the author's claim, I'm willing to die on the following hills:
1. Kumon sheets are the antithesis to creativity 2. Understanding is not a form of memorization (not the rote variety anyway)
I've thought a lot about education, and my personal take is that in the US we way undervalue drilling (by which I mainly mean building up familiarity and muscle memory) and way overvalue understanding.
I've been collecting quotes about these topics for a few years. One relevant to creativity and drilling is Bob Dylan's
> If you sang "John Henry" as many times as me.... you'd have written "How many roads must a man walk down?" too.
There is definitely a lot of value in practice and repetition. I don't think rote memorization / drilling are the only means of getting that practice and repetition. Ironically, with a bit of creativity, we can provide both. Lot of practice, lot of repetition, paired with understanding, play and making things.
Creativity is actually defined by this transformation of finding connections between raw data that you already have to know.
Consider use cases for a rock.
Boring would be using it as a paperweight or throwing it through a window.
Novel but uncreative would be throwing it at the sun, or painting it red. Novel, but kind of useless.
But what about using a rock to play rock paper scissors? Planting it in the soil and watching it grow? That's kind of novel, by way of subverting rock's rules (it doesn't grow, unlike plants) or transforming the concept of 'rock' itself — a real rock isn't needed for rock paper scissors.
So only connections between known concepts are creative. Others might be novel, but useless.
Developing heuristics to categorize patterns and internalize concepts != memorization. If anything, it is the opposite of memorization.
Completely -- it's sort of glanced over in my post as an intermediary step to get the next "thing to be memorized" for two reasons:
1) I've often found these heuristics from books/online/mentors and just had to memorize instead of create them
2) In my own experience heuristic creation has been less of a bottleneck than committing to memory
But it is certainly a key piece of info to have
Being forced to do rote exercises sometimes makes you creative. Solve a thousand trivial multiplication problems and you will spontaneously discover lots of shortcuts, patterns, intuitions that can warn you when you make a mistake and so on.
A common issue I notice when people discuss the terrible state of math education in the US is that teachers demand that you solve a problem a specific way, such as multiplying two-digit numbers by drawing base-ten blocks and applying the distributive property.
People who are good at doing multiplication in their head think the method makes perfect sense and don't know what all the fuss is about. But I believe that those people learned how to apply the distributive property "by themselves". That is, by adults forcing them to multiply two-digit numbers over and over until they developed an intuition of the distributive property by necessity.
When people who didn't go through countless drills are taught the base-ten method directly, they have a harder time understanding it. So ironically it is the students who "mindlessly" drill trivial computations over and over that are more prepared to have a "true" understanding of the distributive property, while the ones whose teachers believe drilling is for chumps and try to just explicitly show them the true distributive right away, they end up memorizing the words of the distributive property without understanding it.
I think this analysis as a bit guilty of over-fitting -- it is quite easy to rote memorize a bunch of things while having little to no understanding of what they are or how they work. Trivial examples include training a room of people to memorize a series of facts written in a language they don't speak (the fact that they have memorized doesn't at all mean they have any understanding of the content). So I would say it's not memorization per say, but meaningful exposure to a thing, the more chances you have to meaningfully interact with a thing, the higher the chance is you will learn how to manipulate it and do things with it. This is the difference between understanding and mere memorization, and the more exposure you have, the greater the chance you will start to see the patterns and understand, versus focusing your efforts on memorizing which will just lead to over-fitting and not understanding. As with NNs, so with humans.
On the other hand, episodic memory (insofar as that is distinct from "3*9=27" memorization) is built on top of creativity.
The vast majority of what we consider "memories" are the creative brain doing an on-the-fly story generation, massaged until it "seems right" and serviced plus a big dollop of emotional confidence.
I hear with llm/chatGPT people don't have to blog anymore, but rather the generated plausible-sounding well-structured nonsense flows like an open sewer onto the web.
1. Creativity in a commercial context once stolen/cloned through back-channels accrues value, and manifests as several competitive campaigns
2. New disruptive ideas are usually shelved until the IP/patents expire. No one wants to go through the sometimes impossible licensing process
3. Emerging technology is usually degraded in the rush for IP assets by established firms i.e. large firms dump billions on ridiculous concepts out of fear of market fragmentation
4. Startups do not usually have cash to burn on speculative IP. Thus, real cutting-edge experimental technology is sometimes never made public for numerous reasons.
Creativity:
https://www.youtube.com/watch?v=VUVix0STUqo
Best of luck, =3
I disagree with elements of this article, but enjoyed reading it. True creatives do not subvert norms consciously or with an acute awareness as part of this article suggests, I agree they need to be exposed to the norms to generate their own interpretation, but I don’t believe that true creativity is a conscious exercise.
Completely agree -- but isn't internalization of their art (i.e. memorization) needed to achieve this subconscious creativity?
That's the point I'm trying to make at least -- that unintentional creativity stems from learning, which at it's core is memorization
Well said - what's funny is that many creative ideas (from an outside perspective) are often very simple connections of two existing ideas for the person that actually did something creative! It's just that other people don't have the same context/knowledge as the creator, so an idea can seem extremely original to them
I agree and feel likewise the author's comments on the above about memorization, but I realized - A higher framework
and is as below, https://rajivkapur.com/3-pillars-of-vedantic-practices/
You have to know things to be reasonably creative but there is a point where memorizing more stifles creativity. Memorization is often very passive and is fundamentally different from searching for new ways of doing things. It's hard to make sweeping statements like this because there are different modes of memory and different modes of creativity.
There is a lot of memory involved in being creative, but I think setting out to memorize things is a bad way to be creative. You have to practice being creative. In doing so, you will naturally remember a lot of stuff like what works, what doesn't, and most importantly which types of things you ought to memorize. For example if you're programming you will find it useful to remember the syntax of your languages. If you're writing you'll find it useful to remember styles and vocabulary. And so on...
Agreed -- I'm not trying to suggest that memory is a replacement for creativity
I'm suggesting that it enables it, as it's hard to be creative when you're still trying to remember the basics of your art
But once things become autonomous -- you can focus on those higher-level explorations
Good discussion on this article. I went to a Vampire Weekend concert last night and they used the 45 minute encore to play covers that were requested by the audience. They got through about 10 songs (a verse or two and chorus), everything from Talking Heads, to Creed, Sublime, Beastie Boys, and Prince and Bob Dylan (the concert was in Minneapolis, so they probably rehearsed the last two). So basically the band knew/memorized the tune for hundreds/thousands of songs from the past couple decades and the singer knew/memorized the words. So I think that this example supports the authors premise that creativity comes from some form of memorization.
Regarding Asian cultures and memorization, there is a recent change that has to be pointed out. The role of memorization classically was to aid manana/contemplation after shravana/hearing-the-teaching. The stage of manana is a kind of inquiry where you raise questions and clarify them or see how the teaching applies to various situations in your life(Like learning a principle in physics and solving problems based on the principle). In the case of poetic works the inner-feeling/bhaava sinks into you allowing one to access blissful states.
However, in the modern context, this has been transformed into memorizing arcane lists/tables, on which one is graded without any further manana.
That thesis is way, too oversimplified and a bit misleading. It could lead you to think the real source of creativity is how much you store, facts or patterns. Put Wikipedia on a computer with a pile of heuristics (eg How to Solve It). Then, it will be more creative than humans in no time. Yet, humans with virtually no knowledge are more creative.
My first study of creativity was Cracking Creativity by Michalko. Skimming its tools, I noticed all of them were about changing how one looks at a problem or connects its pieces. Another work told me geniuses are defined by what they forget or ignore more than what they memorize. The two ideas combined into my working approach.
When I did CompSci research, I would look through the papers for the concepts they reported on. High level ones, core prerequisites, and techniques in how they are combined. From there, I could enumerate variations on each. Then, just keep combining them in straight-forward or random ways. Eventually, something emerges.
Likewise, prior work on creativity and A.I. showed reasoning by analogy was huge in humans. We find patterns in one domain that we generalize to look for patterns in another domain. Then, there’s some process of knowing when to try one or not.
These processes so far are extracting a tiny amount of information, filtering most of it, identifying heuristics in a domain, and heuristics across domains. Also, letting the mind just soak on things to do whatever it does in the middle of the night. These are collectively creativity.
Memorization is a building block of, but different from, creativity. The proof is how we’ve long had memory and reasoning in systems but they sucked at creativity. Recently, systems are extrapolating enough to be more creative but are hallucinating nonsense they definitely didn’t remember. So, they’re orthogonal.
“Creativity comes to those who have internalized the patterns of their art -- they can see the connection or novelty because it's all in their head.”
Edit: I’ll also add that, for planning and creativity, many of us use a deliberate, trial-and-error process that takes time. We don’t just “see” it from something we internalized. We work toward it using the creative process. We usually do see it when it’s finished, though.
The acts of the mind, where in it exerts its power over simple ideas, are chiefly these three:
1. Combining several simple ideas into one compound one, and thus all complex ideas are made.
2. The second is bringing two ideas, whether simple or complex, together, and setting them by one another so as to take a view of them at once, without uniting them into one, by which it gets all its ideas of relations.
3. The third is separating them from all other ideas that accompany them in their real existence: this is called abstraction, and thus all its general ideas are made.
John Locke, An Essay Concerning Human Understanding (1690)
It’s not memorizing, it’s actually knowing and understanding the utility of different concepts. When you learn of a problem in a new field, you go through your bucket of tools and modify one to fit the new problem.
Exactly. And "synthesis" is the better word than "creativity". I wrote an essay about this very topic some years ago: https://docs.google.com/document/d/1tbMTpkPWkkN8_KH2KSpteFsb.... (if you can't open Google docs, an older draft is here: https://rogerwagner.com/creativity.html)
Interesting -- hadn't heard of this synthesis vs creativity take before
I could see that -- agreed that "creativity" can be too blunt of a word to use for all situations
Thanks for reading!
Isn't learning which tools in the bucket fits the a problem best just memorizing a heuristic?
That's the point I'm attempting to make, that it's not blind memorization without context, but still memorization of a heuristic at its core
Probably you would have had more agreement, if you would have used a different word (though I’m not able to suggest that different word that captures the meaning I think you intended)
For me anyway, it's getting so familiar with something that your mind wanders and considers other possibilities.
I'm not in anyway an expert, so I googled what some research says. Here's an interesting meta-analysis (https://link.springer.com/article/10.3758/s13423-023-02303-4). Memory and creativity are a lot more complex than I realized. There are different types of each, and it seems like they interact in complex ways. Here's the findings from the abstract:
> We found a small but significant (r = .19) correlation between memory and creative cognition. Among semantic, episodic, working, and short-term memory, all correlations were significant, but semantic memory – particularly verbal fluency, the ability to strategically retrieve information from long-term memory – was found to drive this relationship. Further, working memory capacity was found to be more strongly related to convergent than divergent creative thinking. We also found that within visual creativity, the relationship with visual memory was greater than that of verbal memory, but within verbal creativity, the relationship with verbal memory was greater than that of visual memory. Finally, the memory- creativity correlation was larger for children compared to young adults despite no impact of age on the overall effect size. These results yield three key conclusions: (1) semantic memory supports both verbal and nonverbal creative thinking, (2) working memory supports convergent creative thinking, and (3) the cognitive control of memory is central to performance on creative thinking tasks.
So some memory seems to be correlated with convergent creativity, which according to wikipedia (https://en.wikipedia.org/wiki/Convergent_thinking) is "the ability to give the 'correct' answer to questions that do not require novel ideas, for instance on standardized multiple-choice tests for intelligence." It sounds like there's less correlation with divergent creativity, which (again from wikipedia (https://en.wikipedia.org/wiki/Divergent_thinking)) is "a thought process used to generate creative ideas by exploring many possible solutions."
But my real takeaway is that people here seem to have strong (emotional?) opinions on "memorization vs creativity: which is better", but few people seemed to bother reading page 1 google results on the topic. So I like to think that bothering to do some cursory research beats both. :)
I see it like this: after I have seen some colors and a ball I can then imagine a ball in any of those colors.
In other words, everything I have experienced and memorized becomes this pool of resources I can imagine from. More I have seen, more combinations I can imagine.
Then I understand how traveling actually broadens my view. It’s not just some nice phrase but hard reality.
Also this means maybe anything we can imagine we can also create. Because whatever I can imagine I can also plot a path from here to there, imagine all the steps in-between.
You may as well say that creativity comes from writing. Because obviously all of the most creative writers write. And the writings of creative non-writers are entirely absent.
I agree with the conclusions of the article. A concert pianist can only add his/her creativity to the piece if he/she totally internalized the piece.
The category is not the thing categorised.
Just because you can put things into boxes does not mean that everything belongs in a box. Whatever essential element you seek to create, whether or not it is concrete or abstract, can simply be put forward as a target of memorisation, without pausing to think about whether you can truly memorise it.
E.g. a heuristic for determining the best heuristic. Simple, just memorise it, right?
"Creativity is just connecting things. When you ask creative people how they did something, they feel a little guilty because they didn't really do it, they just saw something. It seemed obvious to them after a while. That's because they were able to connect experiences they've had and synthesize new things." - Steve Jobs
Something I noticed from being raised by Indian parents while going through the US school system
And again after learning how to acquire new skills quickly
I will disagree. Creativity comes from applying acquired knowledge (that's where memorization comes into account) in new contexts.
Appreciate you reading!
But how do you know how to apply this acquired knowledge in this new context?
It's some form of pattern matching right -- which imo is just a less obvious form of memorization
i.e. you've memorized the match between inherent traits of the context with a specific application of that knowledge
> It's some form of pattern matching right -- which imo is just a less obvious form of memorization
I don't think that meets the commonly accepted definition of memorization.
E.g. https://dictionary.cambridge.org/dictionary/english/memoriza... Memorization: the act or process of learning something so that you will remember it exactly.
If you are transforming the knowledge, i dont think that is the same as memorizing it. In fact, i think most people would describe that as the opposite of memorization.
Using your definition, everything would be memorization. For example, you could describe picasso as simply pattern matching on other paintings to make something new, but i think it would be crazy to describe his work as an act of memorization (then again, the way AI generated art works... maybe its not so crazy)
> But how do you know how to apply this acquired knowledge in this new context?
That question has a false precondition baked in. If you know how, it's not creative.
> It's some form of pattern matching right -- which imo is just a less obvious form of memorization
No. Sensing and matching patterns does not imply memorization. Everything you're saying is completely loaded.
Did you just discover memorization? Because the pattern I see in your words is similar to anyone who's just learned a new tool or technique - they overapply it everywhere as they learn to use it.
To me, “memorization” implies an active process focused on learning a particular set of “matches” (to adopt the term you’re using here). But it seems to me that tacit knowledge (and other products of less concentrated/deliberate learning) often plays a substantial role in creativity.
That is, creativity fundamentally comes from internalized knowledge (as the article says) but internalized knowledge doesn’t necessarily come from memorization.
I think this is right, and one implication is that 99.9% of the self-help content you passively consume on the internet will have no impact on your life, because (barring deliberate learning / repetition), you won't remember it in the high-leverage moments when it might have made a difference.
> you won't remember it in the high-leverage moments when it might have made a difference
But how would you learn to remember the right advice in the right context? The act of remembering something when you need it requires you to do more than just memorize it, you have to properly learn when the thing is actually useful and train your mind to recall this advice in those situations.
It is much better to know that an advice exist and learn when to look it up than to memorize the advice without the ability to realize when you would need it.
Thanks for reading!
This brings up an interesting point -- while I think many people read and forget self-help books without ever improving their lives, the way they can mostly help people (imo) is by:
1) identifying a heuristic 2) making memorization easier through stories
The stories, analogies, acronyms they teach all just make memorizing/remembering/learning their heuristic easier
This claim makes little sense because it fails to distinguish between memory and memorization. I memorize almost nothing, but I remember the broad strokes of a lot of things. This allows me to be creative.
In a way, memorization is a severe risk: if you memorized something before it changed, for example, your creativity may not mean much.
Hey thanks for reading!
What I was trying to convey is that fundamentally learning is memorization, whether conscious "rote memorization" or more less-intentional committing to memory from doing an activity
And that recognizing this allows us to speed up the process of learning fundamentals
Which in turn enables creativity as most people see it
Conscious memorization and rote memorization are two different things. I can intentionally put things into memory without doing rote memorization - or even intentionally avoiding rote memorization. The techniques you use also give you different results in terms of whether or how you use the memorized concept or word in foreign language. (For example getting the effect where you can translate a word between foreign and your language, but can not use it foreign language sentence and do not understand it in context without translating.)
repetitio mater studiorum est -- repetition is the mother of learning
"Repetition is the mother of learning, the father of action, which makes it the architect of accomplishment."
You state that "learning is memorization" but I don't think that is true. Of course learning involves something being persisted in one's brain, but stating that memorization and brain persistence persisting are synonyms seems like an incorrect description.
This is like saying that programming competence comes from breathing air.
Creativity (measured by Openness in the Big Five) is a fundamental personality component that cannot be altered, in the same way that your working memory and IQ cant be altered (unsurprisingly they are closely related)
LSD famously increases openness.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3537171/
Also, openness is correlated but not equivalent to creativity.
Related discussion: Variability, not repetition, is the key to mastery - https://news.ycombinator.com/item?id=33354355
Creativity is probably some combination of memorization + randomization.
I heard this idea recently: creativity needs shallow knowledge over a vast range of concepts. It’s probably not “just” that, but I can see how the vast range helps.