Kottke's one of the very few blogs I've been reading for what seems like forever. He's still doing good work. Recently he's been trying membership, which I've not seen on a site like his anywhere else.
In job interviews I always zone in on the most complicated thing someone has worked on and then ask them to explain it to me. Often a thesis or a project or a large system or something low level to do with OS features etc...
It is amazing how rarely people can get it across to me in basic terms. In fact even the idea of breaking it down into non technical concepts seems to be surprising and alien to many people.
This has happened to me several times (not always in interviews). Especially when someone asks about machine learning and I start the explanation with regression, they get pissed. Nothing shows that you are trying to jump on a bandwagon than expecting needless complexity in explanations.
It's only one way to approach things. Many good people are not great explainers. However, I do find a willingness to try and get down to first principles a good sign. I only take it badly when people seem unable to think outside their expertise, or seem unaware of the layers above and beneath the particular strata they work within.
The problem I find with this outlook is that it takes technical terms being jargon at face value. If you can't fall back on progress in language which allows you to express more complex ideas, you're not going to reach the same depth of understanding in simpler words without it taking a lot longer anyways - with most of your time spent reestablishing what you just threw out the window.
If you rehash it in smaller words, just by information density alone, aren't you guaranteed to be losing some detail?
ivanbakel is not making some blanket claim that things always require complex technical terms. He is just saying that language -- including field-specific jargon -- evolves so that people steeped in the language can explain things to each other.
The clearest example of that how hard it is to decypher popular articles about quantum mechanics. Here, even when explaining concepts at a skin deep level, stories they tell are utterly inadequate if not misleading.
That's because even the basic concepts of QM are ones from linear algebra. And is shy Feynman was a sometimes devotee of the "shut up and calculate" school of interpretation.
> If you rehash it in smaller words, just by information density alone, aren't you guaranteed to be losing some detail?
Of course, but when explaining something to another person, one should strive to optimize for being understood, not for the highest information density.
If you skip out the part of the analogy where you actually return to the complex terminology you're simplifying, you haven't produced understanding - you've only made a just-so story which is a reflection of the reality, not a justification for it.
Great epistemological question for the ages. Vectors for the response include audience type, previous knowledge context, goal for how much needs to be conveyed...
This is the key. The person you explained it to could probably not turn around and explain it to someone else, but could walk away thinking that they understood it.
The same thing that it means to hear a bad quality 8bit, 20kbps copy of a song. It's still better than NOT hearing it, and it gives you an idea about what it entails, even if you don't hear all the information in it.
Even in everyday language we make the distinction about understanding something fully and having a rough understanding of it.
It's the other way around: First was the concept, then the agreement which words we use to describe them.
> If you skip out the part of the analogy where you actually return to the complex terminology you're simplifying, you haven't produced understanding
Billions of people know what a telephone does and how to use one without understanding τῆλε ("tele-", distant) and φωνή ("phone", voice, talk, language). They all get the concept and nobody would argue that they need to understand the terminology as well.
Or: I can teach you to play chess without teaching a single chess-specific term. Yes, other players will laugh at you when you call the pawns "smurfs", but when you understand how to handle your smurf-army, opponents will stop laughing after hearing you announcing "death by smurf-supported horse in 2 turns". (Though they'll definitely hate you for calling a knight a horse ;).
> I can teach you to play chess without teaching a single chess-specific term. Yes, other players will laugh at you when you call the pawns "smurfs", but when you understand how to handle your smurf-army
But that doesn't remove the terminology, it only changes it. You can understand chess equally well in English, French, or Lebanese, because the syntax isn't as important as the semantics. You can't teach me chess without talking about the idea of a "piece", no matter what you end up calling it. There has to be some common groundwork.
>Billions of people know what a telephone does and how to use one without understanding τῆλε ("tele-", distant) and φωνή ("phone", voice, talk, language). They all get the concept and nobody would argue that they need to understand the terminology as well.
The etymology of words tends to have no bearing on understanding them technically. To me, that's a non-comparison.
What's really meant is more that plenty of people use a telephone without a comprehensive understanding of EM radiation. That doesn't mean they can't use their phones, it just means if you explain EM radiation to them in terms of phones, they probably aren't going to come away understanding it any better.
First it was that, to understand, we have to "actually return to the complex terminology you're simplifying". I disagree.
Now it is, that we need some common way of communicating. I agree.
But that are two different questions.
> [explaining EM radition in terms of phones does not help]
Nor does explaining EM radiation in terms of oranges. But I'd do neither. I'd first try to figure out whether you struggle with the concept of what a phone is or with the question how it works. Because step 1 in being understood is to understand what you are trying to say, to define the metric you can later use to evaluate your success in your attempt at being understood.
Norvig is a genius, and his genius is reflected in how he can so succinctly and clearly explain how to write a spell checker, without falling into elaborate jargon or droning on without getting to the point.
And that goes equally for his code and his prose, by the way.
There are many people who use a lot more jargon and verbosity to make their points, who are not in the same league as Norvig in terms of intelligence.
But explaining something in simple terms is a necessary step to teaching the concept. Only once the concept is understood can you put a name to it and keep climbing the ladder of abstraction. This article doesn't devalue proper terminology, it values fluency with abstractions.
This. I tend to think a key to really understanding something, is the ability to put that "something" in to contexts at a wide variety of abstraction levels.
That's what I hated the most with uni: Some professors could write 10 blackboards full of formulae, but could hardly put together a few words to explain what we were actually doing. Unsurprisingly, I don't remember much from these lectures. Fortunately that is not all professors, but it worries me how large a percent of them were like this.
Maybe a refinement of this approach is to FIRST explain in simple terms to establish basic understanding and SECOND going deeper with technical jargon.
"Education is a series of small lies." Spoken by my intro to computer engineering professor, upon telling us that most real processors actually have some ternary logic.
The notion here is that educating beginners about a subject is different from communication between experts. Yes, experts use jargon because it is more efficient. But, the notion here is that if you truly understand something, you should be able to find a way to explain it to a non-expert, building up from words and concepts they already understand. And, further, if you can't do that, then perhaps your understanding is not as deep as you thought it was.
> most real processors actually have some ternary logic
Was he referring to the Z state? If so, it's not really ternary logic but rather a "no result" state for cases where there is a control signal involved (e.g., in a tri-state buffer).
Not really, no. More like an unnecessary detail, especially if it was an introductory course on the subject. It's not like knowledge of the Z state will suddenly change how you would design systems.
Modern CPUs have phase-locked loops generating clocks for each region of the chip. These use separate pull-up and pull-down mosfets to control the frequency, so arguably ternary.
Even more non-binary things happen inside DRAM and flash memory.
If something is too complex to be explained to a college freshman, it might be time for a paradigm shift to simplify (I.e. Ptolemaic astronomy vs heliocentric)
Algebra and calculus were one bleeding edge research that only the elite understood, now kids are learning it in middle school.
> If you rehash it in smaller words, just by information density alone, aren't you guaranteed to be losing some detail?
Yes, but knowing which information you can lose is why it is said that you don't understand it until you can explain it. Explaining it in simple terms forces you to over-simplify, and then you have to pick which information is important and which isn't...and doing that requires a deep understanding of the subject (and your audience).
Well, yeah, but the technical detail doesn't have to remain low. When you first explain something to someone you keep it simple so that they get the broad picture, you use analogies and what not. As they learn you can start introducing more and more detail.
It's about explaining in a way the recipient can understand so "simple terms" becomes "terms your recipient can understand".
This is generally true but I will add one wrinkle.
And there are three kinds of explanation:
1. visual
2. mathematical
3. linguistic
So sometimes, you understand something visually, or mathematically, but you are forced to put it into verbal terms (say, over a text only channel, or voice), and then you may seem not to be able to explain it even though you understand it.
I love this distinction. I've heard the headline stated before, and even if not directed at me, I find the very idea irritating and even a bit offensive. It's not necessarily that I don't understand it, it's that I can't prove to you that I understand it.
Next time I see this concept in action, maybe I'll suggest alternate methods of "explaining".
Filed under "I believe, but cannot prove": The three-way tension is a recurring pattern.
When trying to understand new things, I often try to reframe things as a triangle. When pondering intractable problems, I favor three-way solutions. Some quick examples from memory...
Project management: time, money, scope.
aka Quality: Fast, good, cheap.
US Govt balance of power: Executive, Congress, Judiciary
Language design: imparative, declarative, functional
(A name for the three-way tension you will find a lot: trilemma.)
If you have to budget something, it becomes a soft trilemma: I have 24 hours for work + leisure + sleep, so any given combination will be a point inside a triangle whose vertices are (0, work), (0,leisure), (0, sleep). This structure is called a (2-)simplex.
BTW: a regular dilemma (should I spend or save) is a 1-simplex, and is a simple linear combination asave + (1-a)spend.
Now, this is fun for two reasons:
- You can have (n)-lemmas (i.e. n-fold tradeoff structures) that are modeled as (n-1)-simplices. Actually useful: you can do statistical inference on simplices using the Dirichlet distribution.
- A simplicial complex (basically, a set of simplices) can be used to build topological spaces part by part. This kind of maths (algebraic topology) is a whole "south part of the mountain" climb towards abstract mathematics that bypasses a lot of Cantorian handwringing on the ultra-local structure of topology. Instead, you're computing stuff from the get go -- and indeed one of the emerging machine learning techniques goes precisely from building a simplicial-like complex from data and computing characteristics of its topology.
And in my experience, the harder the subject, the more informally experts speak. Partly, I think, because they have less to prove, and partly because the harder the ideas you're talking about, the less you can afford to let language get in the way.
Informal language is the athletic clothing of ideas.
Eh, I suspect part of the reason is that once a topic crosses a certain level of difficulty, trying to convey an exact explanation for it in a reasonable amount of time becomes impossible and experts just fall back on analogies and hand-waving.
I'm a physicist, and if you asked me to explain Newton's laws or classical thermodynamics to someone, I could probably give them a reasonably complete overview in a few hours. And in doing so I'd probably introduce and explain some jargon (entropy, derivatives, etc) along the way to make it easier.
On the otherhand, if you asked me to explain Quantum Field Theory or General Relativity in the same time, I'd basically just give up and throw out some friendly but inexact analogies about stretched rubber sheets and the like, and hope the walked away with some vague sense of the outlines of the subject.
Meh, that's a cop-out. "You don't need complex sentences to express complex ideas." and "They don't use more complex sentences, just different words"? It's pretending that the complexity in the language lives only in the grammar, and not the jargon. Think of it this way: if spoken language gets more informal in terms of sentence structure the harder the topic, that means that the easier the topic, the more complex the sentences.
I think PG has confused the informality people have when they have a common understanding with something else. Do you really think that a pair of experienced mechanics in an auto shop discussing a basic repair are going to be using more formal language? Turn one of the mechanics into an apprentice and consider the same discussion.
(Highly recommend the whole Feynman series, which is excerpted from an interview video called The Pleasure of Finding Things Out, which is also spectacular)
Some parts of physics might stretch the ELI5 to its limits. I like to explain science to my non-science friends, but I'm always hitting the issue that parts of the specialized knowledge become intuitive, and then it's difficult to go back. Say, my understanding of how electrons behave in matter is a category on its own (let's call it quantum), and it's difficult to translate to common terms.
The more I try to learn physics, the more I need to see the mathematics to understand the ideas. Without the math, I don't believe it is possible to break beyond the surface 'armchair' level of understanding.
I started by building an intuitive understanding of advanced physics before I really dug into the math. It is absolutely possible. That being said, I did have a very good understanding of advanced mathematical concepts to start with, so I could think intuitively about things like high-dimensional spaces and eigenfunctions.
That's because those things are inherently tied. It's not about taking your time and making the layman an expert to understand some hard topic. It's about being able to relate it to them in simple terms.
Obviously, if you use the "hairy" physics etc terms, then the layman wont understand your explanation, because he is not familiar with them.
Unfortunately, that is what is expected more commonly these days where the audience being presented to a.k.a layman aren't users or domain experts or people who know whats going on. They're management "experts" tasked with getting "resources" to "get things done".
I've seen people from my domain making ridiculous, inaccurate analogies and oversimplify to hide the inherent challenges that a task usually involves. And they get away with that because they aren't the ones actually building the product. They just have to sell it. And the one's they are selling to aren't/won't be using the product either, they will tell someone else to use it.
A math formula is one of the objectively simplest ways of expressing something. But it's not easily understood unless you know the math well, thus violating Gruber's rule. Feynman was honest about this, because magnetism can be explained very simply, but not in any familiar way to things most people already know about.
At least as good as math (and capturing the objective complexity better I think) is a working program. "What I can't program, I don't understand."
When people ask for a simple explanation, they usually expect it to be easy for them too, because we all want simple and easy at the same time, even though only one of those is objective. If your reply to "how do magnets work?" is to start by writing down Maxwell's equations, you're gonna get crap for it, but someone who uses the fake rubber band analogy will be well received. But who really understands magnetism better?
But I think that's the deeper point of the "if you can't explain it, you don't understand it": Sure, you can write down the equations, but then what exactly have you done? You wrote down some string of characters and showed that it can be derived from some other strings of characters by applying a series of arbitrary-looking rules. By itself, that doesn't tell anything. It's only when you manage the connect formula to the part of reality that it tries to describe - and connect that part to the rest of reality - that it gains any sense.
Trying to explain a concept in "simple terms" forces you to view it in terms of its connection to other, well-known phenomena.
> You wrote down some string of characters and showed that it can be derived from some other strings of characters by applying a series of arbitrary-looking rules.
That's what it looks like if you only look at the formula. (And why I'd say that formulas are not an "objectively simple way of expressing something*)
If you can somehow keep in mind what the formula is supposed to represent, the operations on characters will let you find some insight about that thing and will stop standing only for itself.
The point is that the term "simple" is inherently subjective - there are plenty of concepts that are simple and intuitive once learned, but are hard to learn.
Also, mathematical formulas are essentially a different language. If people don't speak it, then it's a moot point.
I'd say we do it quite well, just not in in higher brain parts. We find it hard to integrate 'intuition' with the later 'reasoning' parts, instead, it just biases the logic.
Well an equation is certainly the more precise way of describing something, but not really the simplest. My most vivid memory of this is my Automata class; the instructor led with the equations before describing in plain language. Take push down automata as an example [1]. The equations are found in the formal definition, but the informal definition is much clearer.
An equation is compressed, which helps give it its precision, and can create an illusion of simplicity. Just like "a witch did it" is compressed, but if you expand it all of the complexity is inside "witch" and "it", and as you expand "it" you might arrive at some true simplifications in the untwisting sense like physical laws that make you wonder about the need for "witch"... But even if you expand the math it's still all together one of the simplest ways to describe something. That's why I like programs more, there's less implicit compression and it's easier to expand things you aren't familiar with. And there's the formalization of Kolmogorov complexity you can use for things.
As you say, informal explanations are great for understanding and bringing clarity, a lot of the time they're just used to help expand the unfamiliar math, but also a lot of the time they trade off precision, conciseness, or accuracy. At worst they complicate things with twists and relations in the informal explanation that don't exist in the real deal. Sometimes that might be helpful in the same way technically unrelated mnemonics are but it's important to point them out.
The conflation of simple, easy, concise, precise, clear, intuitive.. is the root of the issue when discussing what makes a good explanation.
It's not just some hard physics that have trouble with the ELI5 / "should be able to explain it" concept.
This is also true for literature, poetry, non-analytical philosophy etc. And there the challenges are not about the complexity of the subject, but about its subtlety, and how one needs historical context and/or certain life experiences (or even proclivities) to be able to grasp a particular piece of work.
It's not so much life experiences, but prior knowledge and familiarity.
Taking mathematics as the operative example. A lot of deep theorems are trivial when you fully understand all relevant definitions. These proves cannot be stated 'in simple terms' because it took a stack of non-simple definitions to even formulate such a theorem.
Explaining the entirety of the stack gets hard when the stack gets deep. For example, a lot of math gets really hard to explain when the concept of the real numbers being 'uncountably infinite' whilst the rationals are 'countably infinite' isn't obvious.
Neither if he was speaking about Quantum Mechanic. Feynman probably paraphrasing Bohr:
"If you think you understand quantum mechanics, you don't understand quantum mechanics."
In one of his books he told this story as follows, a cab driver watched him on TV and said to him "I cant believe you actually tried to explain what did you found." Then Feynman said "Well, what would you say?" Cab driver "If i could explain it in 2 min. then it would not be worthy." Afterwards Feynman used that line. I might be wrong though, but pretty sure he told a story like this.
I run into this all of the time when explaining what I'm doing at work as a programmer. It's not that I couldn't explain why my merging workload increased due to the increased code collisions from my coworkers working in the same files, and how dividing my current assignment in two so I could check in code would reduce this. It's doing that in a normal conversation under time and attention constraints which is the problem.
Well he made that statement in 1965, and then wrote QED, which is a layman's explanation of his Nobel Prize work, in 1985. So perhaps he set out to prove himself wrong.
QED does explain quantum electrodynamics without any calculus or other math (for example, he uses imaginary clocks as a way to explain wave interference), but I think if the average person read it they wouldn't walk away with even a partial understanding of it.
I don't see this as evidence the other statement is false. If it was something he understood well enough at the time he was earning the prize, it wouldn't have been groundbreaking work worthy of the prize. Gaining that understanding or paving the way for that understanding is what's prize-worthy.
If one assumes that things at the edge of human understanding and thus most worthy of such a prize are understood only incompletely by those who understand them best, it is both consistent with Feynman's maxim and not at all a sign of hubris.
They are both right. Something's can't be dumbed down enough to explain to everyone, but you can always explain the concepts. An example could be comparing an interface to a chocolate cookie recipe, when explaining it to your grandmother.
She'll still have no idea what an interface actually is, but she'll have a good grasp of the concept.
"implies" and "because" are not equivalent. For example, "x is positive implies x^2 is positive" is true but "x is positive because x^2 is positive" is false.
I understand the point of your comment. Just the last sentence ("Not being able to...") would have been enough. But your monkey example is confusing, because "=>" usually means "implies", not "because", and with that meaning there is nothing weird about the last statement.
"if not a, then not b" is equivalent to "if b then a" and is neither equivalent to nor the opposite of "a implies b". I'm having trouble understanding your point under these confusions. Is it that being able to explain is a sufficient but not a necessary condition for understanding?
Point taken. My point is just because I can simply explain it does not imply clear understanding. And not being able to explain isn't a reliable indicator of lack of understanding
I think this idea is basically nonsense. Some things are complicated. To "explain" them in simple terms you necessarily leave out a lot of information. If all that information isn't crucial to the core idea, maybe that's worthwhile. But sometimes that information is crucial.
Some people look at advanced mathematics or physics and wonder why it has to be so complicated and so full of jargon. It's complicated because it is. The jargon, believe it or not, is mostly an attempt to make it easier to communicate. It would be very, very difficult to wade through these ideas without introducing new words with precise definitions.
Then again, John von Neumann said, "In mathematics you don't understand things. You just get used to them." So maybe the title is true for trivial reasons after all.
Remember that repetition is the heart of learning. If you want someone to understand Relativity to the same degree Einstein understood it, you don't start with the math. You start with an analogy (which obviously doesn't match reality). Once they understand that you can add more analogies (maybe about just parts of Relativity). Eventually you build up from parts to the whole and from broad analogies to specific details. Over the years this is happening you will be gradually introducing shortcuts and jargon. But, only a master of the subject can explain a complex topic in simple terms. I can't tell you how many times my children have asked me a question about something I thought I knew, but once I started trying to explain it in terms they would understand I realized that I really didn't understand what I thought I understood. But if I spent a little bit of time reading up on the subject again then I found that I could explain it to them in such a way that they could correctly explain it back to me.
It's not nonsense. You are forgetting a very important detail: "explaining".
He was not talking about writing a research paper on the simplest possible terms. Rather, about "explaining" a subject. When we are explaining, we always have to deliberately omit details to get the basic idea across, so that the other person can form the initial mental image. Sometimes we also have to resort to metaphors and comparisons from entirely different fields, or relate them to real world concepts.
Once that is done, then the lengthily process can begin on improving the fit of that mental image to the subject. This will require more explanations, more focused, in a different level of detail.
Case in point: dimensions higher than three. Most people cannot conceive what the heck a tesseract is, or why it is usually drawn the way it is. But then you can tell them to watch Feymann's Flatland scenario, and they will be able to understand the basic concept.
Are they mathematicians now? Hardly. But now they have some understanding on the subject which they did not have before, even if it is on a very high level. And can try to absorb more details, if so inclined.
Er, they may be left with a warm and fuzzy feeling that they have some "understanding", but let get them to do some multidimensional analysis and see how they do. The meaning of "explaining" seems just as elusive as the "simple explanations" themselves - once you've heard such a "simple explanation" and have a feeling that you now understand more than you do, what is the objective difference between you-before and you-now? It's my experience, and several professors have confided the same, that the "simple explanations" give students a nice feeling that they "get it", up to the point where they actually try to apply the knowledge and fail miserably. Maybe the explanation then wasn't really "simple", but now we're just shifting the goalposts further into epistemology.
I would tentatively agree that there is a skill which allows you to break down complex subjects into digestible chunks, but that usually depends a lot on the student as well as the teacher(e.g. some people are good at following metaphors, others learn very well from repeated examples, etc), but the outright statement in the topic suggests that there is no such thing as irreducible complexity, which seems ludicrous.
> "Er, they may be left with a warm and fuzzy feeling that they have some "understanding", but let get them to do some multidimensional analysis and see how they do."
Nobody's having any delusions that explaining is the same as teaching. You can explain multidimensional analysis simply if you understand the subject fully, however you can't teach something to someone else unless they work with you and try it out themselves.
But if you want to explain to your politicians that hospitals need special machines to do radiation therapy you don't enroll them into nuclear physics 101, they can probably do with a very broad explanation of how radiation works.
I see it as an analogy with how, in maths, to fully understand a complicated concept you must be able to explain it using more and more basic definitions, but that does not mean you have to express everything in basis to fundamental axioms, just that you could.
Maybe explaining a complex concept in simple terms would take you 2 hours, 12 hours or 3 days, but you can do it if you really understand it.
And more! It is possible, it can be done. But it is hard. That is no excuse for not trying though. Big shout out to Bret Victor's work for starting a lot of this, and thanks to Feynman for encouraging and practicing what he teaches.
Edit: apparently, I have been misinformed for a very long time. These are still excellent lectures to watch though!
The Feynman Lectures are now on Youtube[0], and I like to watch them (all of them) every few years. I highly recommend that if you've never seen them, you take some time and watch them- really watch them. Close the other windows, turn your phone to do not disturb, and really watch these masterpieces of education.
These lectures are different from his famous Lectures on Physics. The lectures you linked to were compiled in "Character of Physical Law" and are not related to his Caltech undergraduate courses where he presented his "Lectures on Physics" content.
To be more clear, the lecture series you are talking about is called "The Character of Physical Law" and is not the same content that one would get from reading "The Feynman Lectures on Physics", which is a rigorous course in most of undergraduate physics. But I highly recommend the "Character" series. Just wanted to avoid having people think they're the same thing.
I can simplify how a car engine works, but that doesn't mean I understand how the air to fuel ratio is obtained.
Opposite example: Simplify how walking works, and make sure to include the critical systems such as major muscle groups, stabilizers, vision, inner ear, thigh/knee/pelvis/hip construction, the curved spine and its connection to the head, and blood pressure flow/regulation.
Still, you understand how it works. Is it enough to build one? Of course not.
You'd be missing a lot of experimental data that has been collected for decades. You'd be missing the metallurgy, etc.
The basic principle is very simple. Maybe you could even build a low powered version, given materials and time.
If you insist that understand how to obtain the proper air to fuel ratio is required to understand how an internal combustion engine operates, then I'd say we don't really understand anything at all. Can you mine and refine the ore to create the allows it is made of? Can you program the ignition (or alternatively, build a carburator)?
The funny thing is, I actually can mine ore, smelt it into iron and forge it. That doesn't mean I can explain how that works (monoxides in a fuel binding to oxides in iron ore to separate iron from slag, aligning grain and changing durability by heat treating, etc). Maybe that doesn't matter if all you want to do is produce pig iron, but if your specific engine design requires a specific grade of alloy, now you need to fully understand metallurgy.
Take it a step further and the engine is built, but you need to make a modification (forced induction, race fuel, etc). If you don't want your engine to blow up, now you need to understand air/fuel mixture. Sure, I can do something with a simplified understanding. But the lack of a complete understanding will leave me with a blown-up engine.
Does it follow that if I understand metallurgy and internal combustion engines, but can't simplify it, that I don't understand it? Of course not; the entire premise was that I already understood it. Simplification does not show you understand something. It can only provide a shallow, incomplete understanding.
So you don't need to be able to simplify to understand. You need to understand to understand. You need to simplify to provide a simplistic, shallow view.
Explaining things in simple terms is fine to some extent, but I personally see it shallow if it's just reduced to that.
If someone (including me) don't want to put some brain energy to understand some context then the recepient, not the messenger is a problem imho.
This is Feynman from the introduction to his Feynman Lectures on Physics:
The question, of course, is how well this experiment has succeeded. My own point of view — which, however, does not seem to be shared by most of the people who worked with the students — is pessimistic. I don’t think I did very well by the students. When I look at the way the majority of the students handled the problems on the examinations, I think the system is a failure. Of course, my friends point out to me that there were one or two dozen students who — very surprisingly — understood almost everything in all of the lectures, and who were quite active in working with the material and worrying about the many points in an excited and interested way. These people have now, I believe, a first rate background in physics — and they are, after all, the ones I was trying to get at. But then, “The power of instruction is seldom of much efficacy except in those happy dispositions where it is almost superfluous. ” (Gibbon)
Note that by his own account, most of his students did not do well. James Gleick's biography of Feynman, Genius, has a longer discussion of the disappointing results of his lectures to undergraduates at Caltech, many of whom reportedly stopped attending the lectures as they were not getting anything useful out of them.
That Feynman in fact had difficulty explaining freshman physics to the highly qualified students at Caltech surely does not indicate he did not understand freshman physics.
Some topics are simply very complex. It is not clear that they can always be conveyed in simple terms. In some cases, a "big picture" explanation may be possible but the details remain complicated. In some cases, a hand-waving analogy to some everyday phenomenon may create the illusion of understanding but be misleading or wrong.
To give a specific modern example, a state of the art video codec such as H.264 is extremely complex, built of many complicated components and sub-algorithms. While it may be possible to explain the big picture in relatively simple terms, the detailed implementation and operation is not simple. The inability of someone who creates or implements a video codec to explain it in simple terms to a layman is not an indication that they do not understand it.
The best math professor I ever had, was in a 400-level Real Analysis class. His particular skill was that he basically had mastered the catalog of incorrect mental models students developed.
He would be proving something in class, someone would stop to ask a question, and from their question he would know not to try to answer it as asked, but rather he knew they were asking the wrong question, because the question was predicated on a misunderstanding of something three or four steps back.
Now these were proofs that a working mathematician could probably do in their sleep, but they were a level up in terms of abstraction for most of the students and to be able to drag us along with him was a rare talent.
I always felt that the Feynman Lectures were really, "an incredibly deep understanding of freshman level physics, targeted towards physics graduate students." (As opposed to freshman level physics targeted towards pre-med students.)
Even after I got my undergrad degree in physics I would reread the Lectures and would see things on a whole new level than I ever understood them the first time.
Edit: even "Surely you're Joking" had jokes that only someone that had graduate level general relativity would get (talking about "gee-mue-nue, gee-mue-nue")
We now have programs that "understand" certain domains better than humans. And yet there are no "simple terms" to explain their understanding, reduce it to a neat closed-form formula.
Isn't it a mistake to assume a nice, elegant explanation, a cute human "narrative", must always exist?
"But Radim, the computers don't really understand the domain!". Sorry, a cop out, No True Scotsman fallacy. In what way do they not?
This is crystal clear with Go: the little stories humans have built to deepen their understanding of the game over 2,000 years are irrelevant to a game's outcome. The details and variations are all that matters, and there is no way to reach the computer's level of understanding of the game without understanding the details.
I like the sentiment behind this article, but simplification (like lossy compression) omits information for the cause of simplicity, leading to an incomplete understanding.
Take for example legal concepts like securities law or environmental regulation. Yes, you can "simplify" an explanation of the Securities Act or the Paris Accord enough to fit them into a tweet, but you lose information necessary to formulating a full understanding.
If you're trying to have an informed debate about policy adoption, the details matter.
Being able to explain something in simple terms is actually REALLY difficult to do. It is a skill in itself. Someone can intuitively understand math very well, but lack the skill to explain it to someone else at all.
I love organizing, simplifying things. Geeks are addicted to complexity. I think it's some kinda machismo thing.
This incompatability has caused me problems my entire career. I've more or less stopped trying. I just do what I'm told and save my creative energy for my personal projects.
If you can explain something in simple terms you may not understand the complex details.
If you hold a complex idea in your head translating that into English can be difficult because part of the process is removing/altering information to fit into existing notions. That is why buzzwords are popular they can take an idea and put it in a relatable concepts for the masses.
This strikes me as raw arrogance. Complexity is intrinsic to many systems that are subject of expert study. To tell someone who has devoted their career to understanding a complex topic that they don't understand their subject because they can't express it in layman's terms without doing terrible violence to the underlying phenomenon is ludicrous.
This is the sort of thing you'd believe if you were an arrogant 20-something who thought they could learn any subject in a few hours, cushioned thoroughly by the illusion of understanding.
"Oh yeah, I understand the mechanisms of human vision. It's just rods and cones."
"I understand the causes of the American revolution. It was just people protecting their property."
"I understand Joyce's Ulysses. It's just follows three people from Dublin over a single day. I read the Cliffs notes."
"I understand why coffee makes me alert. It's just blocking some brain things that make you sleepy."
Now, I will agree that if you don't know how to break interactions down into teachable parts, you will probably have trouble as an engineer or scientist both advancing your own knowledge and introducing people to the field. But to suggest that your understanding of a subject hinges on being able to deliver an explanation in simple terms is just silly.
Complexity is real, but understanding means being able to conceptualize a system at such a high level that it becomes possible to convey the most important key ideas simply. This may involve some amount of over-simplification (which should be indicated if present) and will undoubtedly leave out a tremendous amount of detail, but the conceptual essence should be there.
The inability to do this is IMHO a sign of either lack of true and deep understanding or communication ability.
There are fields where nobody can really do this. These are fields that nobody truly understands yet.
Edit:
Let me give you an example. I had a bad (but typical) math prof in my very first calculus class who covered the board with examples and took us through how to differentiate a function. I was completely confused and stuck. Then I called my father on the phone and asked him and he told me "a derivative is a rate of change." Then I got the rest of calculus.
The prof had talked about slope but that wasn't conceptual enough. I needed "rate of change." That is the essence of calculus. It's the most important concept. Calculus deals with rates of change and the inverse (integrals). Once this concept was communicated the rest of calculus became possible to understand.
If you can't do that for a given area, you don't understand it or (as was the case with this math prof) are a poor communicator.
> This may involve some amount of over-simplification (which should be indicated if present) and will undoubtedly leave out a tremendous amount of detail, but the conceptual essence should be there.
I think the disagreement hinges on whether you believe the snippets from the GP post e.g. the "rods and cones" or caffeine examples are fundamentally valuable or not.
> There are fields where nobody can really do this. These are fields that nobody truly understands yet.
Curious what you would consider an example of such a field?
It's a good book, and I enjoyed reading it. But over subsequent years I've had growing doubts over the true value of the "understanding" that it provided.
> Curious what you would consider an example of such a field?
The frontiers of physics always strike me this way. The names and explanations are incredibly strained and bizarre. This is probably the canonical example and is perhaps why The Elegant Universe doesn't quite achieve what it sets out to achieve.
Another one from my own academic past is "complexity theory" where hand-wavey concepts like "emergency" or "higher order causation" get thrown around with a lot of confusion. We can clearly see there's some "there" here, but I don't think we really understand what it all means yet. Put stuff together, interesting higher-order stuff happens, and so now let's create a combinatorial explosion of jargon around it. Publish or perish.
Finally "consciousness studies" or anything else semi-fringe probably counts. Again there's certainly some "there" there, but we don't even know the right questions to ask. Something is happening to matter that makes it wonder what's happening to it.
For me the "derivative is a rate of change" moment was when I was told that a limited integral is the area under the graph. It just never made sense till then.
I suspect a so-called expert's ability to solve problems within a domain (i.e., how "expert" they are at that domain) would be a function of their ability to manipulate terms in that domain which to me is akin to their ability to those terms clearly.
I can't imagine someone easily solving problems within a domain wherein the terms they use are giant spaghetti balls of complexity. Anything can always be broken down into further levels of abstractions (or prove me wrong on that).
Have you ever 'taught' anyone anything? When you do, you'll probably get what Feynman was trying to say. Complexity is intrinsic. Just that as a teacher/guide/explainer, you have to gently guide the novices around enough to invoke a sense of curiosity that will drive the pupil to explore on their own. The task at hand here is to make things less overwhelming, not removing or ignoring complexity.
For every complex system, there is usually an underlying simpler, though not accurate, model that makes grasping the broader concepts easier.
I think Feynman's statement is a wee bit over the top, and I will say that being able to teach complex things to novices is definitely an art and certainly one which I do not possess
I think it's partly motivational. There is definitely a level of understanding where it feels like you understand a concept, and then when you try to teach it to someone else, you realize when they ask basic questions that you don't understand it as deeply as you thought you did.
So, if you think you understand something, it's worth trying to explain it simply, so you can push your own understanding of the subject.
> For every complex system, there is usually an underlying simpler, though not accurate, model that makes grasping the broader concepts easier
One doesn't stop at that model. (Feynman didn't.)
It's used as scaffolding, later torn down, to organize thoughts. By continuously proceeding downwards, in a fractal nature, filling in gaps, tearing and replacing where needed, the larger structure always stands ready to contextualise.
This helps hold the audience's interest. It also lets the mind, by itself, fill in gaps, promoting retention as well as pedagogical efficiency.
I have, and do. Physics 101, or in my case computer science 101 are not the world. Humans are gifted teachers (probably unique among animals in our instinct to simplify while teaching). I love distilling things down into teachable parts. It's hard, interesting work.
But I also do research, and I absolutely do not expect deeply technical topics like advances in aerospace engineering or gene regulation or cosmology to have "simple explanations" that I can understand. I accept that I won't understand them without years of effort, and I won't denigrate experts in those fields because of my limitations.
I tell people to come up with elevator pitches for complex work all the time, but I don't evaluate the quality of a thinker by the quality of their elevator pitch.
> "But I also do research, and I absolutely do not expect deeply technical topics like advances in aerospace engineering or gene regulation or cosmology to have "simple explanations" that I can understand."
There's a difference between explaining the entire topic and just giving you a very broad strokes approach to get you started learning the subject on your own. Nobody's having any delusions about summarizing a complex subject to a few sentences to substitute learning it the hard way, but if you can't give a very gentle explanation that can ease people in to the subject so they can get started by themselves, I would say that you don't fully understand the subject.
Also, I would add that this explanation is not universal at all. You have to be able to gauge your audience and adapt to what's most suitable.
I agree cs 101 is not the world. I also agree some concepts cannot be simplified and it shouldn't.
My point was what you hinted towards when you say "distilling things down". My gripe is that most people that have taught or guided me have had no inclination to do that. I would have had a much smoother learning curve for it.
I also feel if there is a concept that can be taught by one person to another, it can be slowly be presented in increasing order of complexity than be presented all at once. That feeling of slowly building or adding complexity is underestimated and hard.
I agree cs 101 is not the world. I also agree some concepts cannot be simplified and it shouldn't.
My point was what you hinted towards when you say "distilling things down". My gripe is that most people that have taught or guided me have had no inclination to do that. I would have had a much smoother learning curve for it.
I also feel if there is a concept that can be taught by one person to another, it can be slowly be presented in increasing order of complexity than be presented all at once. That feeling of slowly building or adding complexity is underestimated and hard.
Being able to convey information in simple terms and pass along the depths of your understanding are two independent topics.
As an example, when we teach people how to fly a wingsuit, we essentially teach wingsuiting (which is a massively complex pursuit) as a set of nested arrays of increasing complexity.
For example, for a first jump course, the goals are as follows:
- Exit safely
- Demonstrate ability to navigate in-flight
- Deploy
These are the most foundational aspects of flying a nylon dress out of an airplane.
After a few jumps, we'll add complexity to each of those, so to shift one point, it starts looking like:
- Exit in an unstable manner and gain stability in less than three seconds
In order to do that, you'll need to understand what causes a wingsuit to be stable, why it gets unstable, what happens when it gets unstable, and how to correct it. Additionally, you'll understand why I want you to do it in less than three seconds. However, at first, you just need to get out of the fucking plane. If I try to tell you all this extra shit you won't remember the foundational thing I need you to remember. You'll probably get unstable, you'll probably figure it out kinda, and you'll probably at some point deploy a parachute.
I think it's meant at a higher level. This is one of the basic tenets of philosopher Ludwig Wittgenstein's ideas:
"The limits of one's language are the limits of one's world." It's controversial, but the concept being if you can't speak about it, you really can't know it fully. This gets very fuzzy when experiencing art or emotions, but there is a cold logic to it that I appreciate, too.
I dunno. I learned more about light from [1] than anything else i'd ever read or saw. Am i an expert? hell no. I do have a far, far better handle on how lenses and mirrors work though. I think you'd be hard pressed to claim that either qed isn't that complicated, or, this lecture does a bad job of describing it.
Although, i would agree that some experts do a bad job of explaining things. They're still experts. I think you'd agree that non-experts don't really have a chance at explaining things.
Nothing I do is intrinsically complex. It's just data processing. Copy a string from here, paste it over there.
Maybe, maybe, some times I'll do something clever.
Some times someone will have an insight, intuition. Like a market analysis. Or frame some useful questions for our recommenders.
Otherwise, all of the complexity that I encounter day-to-day is because of people's cognitive limitations. We make it hard. Misunderstanding, miscommunication, separation anxiety, dogma, arrogance, cya, whatever.
? There's a difference between explaining something complex to a layman and explaining something complex in simple terms. You can always do the latter, and if you can't you don't understand the topic; this is a different exercise than trying to fit a topic into someone else's horizon of understanding, which may be thousands of simple building blocks behind.
Understanding the fine details is one thing, but unless you can really communicate a synthesis or an analogy, it's impossible to really prove you understand something.
Nobody cares about a genius who understands something until he can really interact about the subject.
Humans work with communication. When you learn about something, you read about it and you process it. To see if you understand it, you must communicate it back. It's not really possible to guess how you brain just learned about something unless you ask that brain to reformulate it.
I care that the people at CERN can make a working LHC, not that they can communicate it to me succinctly (eg, in under 2 hours).
It's fine that it takes 150-300 hours of back-and-forth communication to transfer the ideas behind how it actually works. That seems like a reasonable amount of time to communicate a complex idea to a novice.
There's also layers of understanding:
The LHC bangs two rocks together hard to see what happens when they break.
The LHC use magnets to accelerate protons fast enough they break apart because it recreates conditions similar to the early universe.
"The LHC use magnets to accelerate protons" - That statement would really confuse someone with a basic knowledge of physics because static magnetic fields can't do work and hence can't accelerate particles.
The acceleration is actually done by electric fields produced in RF cavities.
I don't believe you'll get a better than "smash rocks, see what happens" understanding of what the LHC does without significant time invested in understanding.
That's okay: politicians don't actually need to know how it works, they need to know why they should build it -- which is a) something they have domain expertise in and b) a much easier question.
"To see if you understand it, you must communicate it back"
That's one option. You could also apply the knowledge and see if what you think you understand has resulted in a mental model from which you can make testable predictions that correspond to the real-world behavior of the modeled thing.
I think it's meant to apply to a single, atomic concept that the audience is suitably prepared to understand. It couldn't be something like "how to build a car starting from a lump of steel" -- that would not be a single, atomic idea. "How anti-lock brakes work" might be a better idea to explain in simple terms.
His point seemed to be that experts who don't know how to converse with non-experts aren't really expert enough to hold the title. If you don't know what concepts you had to use to learn the material, then how can you trust your conclusions?
It's sort of like saying that a mathematician who doesn't know how to prove their results to one who only knows the axioms doesn't fully understand their results.
I think you misunderstand what the author means by simple terms.
Everything can be talked about on different levels of abstraction, including ones which use metaphors or contain inaccuracies in order to create a jumping-off point for further learning.
If you understand something very well, you're able to use various levels of precision + abstraction in order to best communicate it to your audience. You can "play" with it, without getting confused or off-track. If you don't understand it well, this gets very hard.
I think this is true for teaching and mastery in general, and not just for teaching about science or complexity.
I think your examples prove the opposite, that if you can explain something in simple terms than you do not necessarily understand it.
I do agree that poor communication skills are not a lack of understanding, however, not being able to explain something I thought I understood is a personal redflag.
from the zen of python:
If the implementation is hard to explain, it's a bad idea.
If the implementation is easy to explain, it may be a good idea.
As an arrogant 20-something year old, I disagree :P.
An important distinction to make here is whether you're teaching someone to intuitively reason about something or to logically calculate it precisely.
I think as long as someone grasps the basics of high school math and has a decent working memory, they should be able to learn how to calculate anything precisely given that you break it down into simple steps. If you don't know how to do that, you have no clue what you're doing.
Then there's intuitive reasoning. I think for most things, if you're familiar with a topic, you should be able to teach it in a way that makes sense. Of course there's exceptions -- some people are incredibly brilliant, but lack social awareness. I think that's the exception rather than the average case.
But not all topics are so easily reduced. There's probably some exceptions. The strongest that comes to mind is quantum computing. As many times as you can explain things like superdense coding, there's still a sense of "magic". Where the results of the math seem unnatural. And you go back through each step and try to figure out where things went weird. But each step is a logical progression from the beginning.
On that note, Michael Nielson's "Quantum Computing for the Determined" is one of the most well-taught courses I've found, and does a very good job of stepping the viewer through a very complicated topic. Michael Nielson clearly knows what he's doing :)
> I think as long as someone grasps the basics of high school math and has a decent working memory, they should be able to learn how to calculate anything precisely given that you break it down into simple steps. If you don't know how to do that, you have no clue what you're doing.
Well, yes, if the only goal is computation. It doesn't take many instructions at all to be Turing-complete, and then you can compute everything a Turing machine can, which is lots of things.
That said, being able to calculate something can be very meaningless. You can try to do AES[1] by hand, if you'd like; it would be frustrating and very time-consuming, and then at the end of it you'll have gained effectively no understanding of its design or structure.
Using AES as a case study as well, one could explain the basic idea of encryption to a layman without too many troubles, but I can't imagine trying to explain anything about its operation or design to a layperson in a way that would make any sense at all.
> To tell someone who has devoted their career to
> understanding a complex topic that they don't
> understand their subject because they can't express
> it in layman's terms without doing terrible
> violence to the underlying phenomenon is ludicrous.
The article doesn't say you should be able to explain it to a layperson, but to a beginning student. This is a distinction that matters. A beginning student in your field should have enough preparation that what you need to explain is only the parts that relate to your particular area of expertise.
No one should devote their career to simply understanding a topic. If all you do is load someone else's work into your head, what value you are adding? A career is a mixture of learning as well as exploring and discovering new things. Explaining just the parts that are known is a much smaller order than transmitting the effort of your entire career into words.
Note also that Feynman isn't claiming you should be doing this all the time, but that you should be able to. I've been programming for about two decades and I've learned a ton (and yet still have even more left to learn). I'd like to believe that, yes, I could explain almost all of it to a beginning CS student or even a lay person.
They might not have the patience for me to build up all of the necessary structure from one simple piece at a time, but I think I could. After, that's how it got into my head in the first place. The parts that I couldn't do a good job decomposing and walking through are exactly the parts that I probably don't have a good handle on. (For me, networking and operating systems come to mind. I know some of the jargon, but I don't really know how it all works.)
236 comments
[ 2.6 ms ] story [ 290 ms ] threadIt is amazing how rarely people can get it across to me in basic terms. In fact even the idea of breaking it down into non technical concepts seems to be surprising and alien to many people.
I really admire those who can.
If you rehash it in smaller words, just by information density alone, aren't you guaranteed to be losing some detail?
Executing on a concept using complex technology requires complex technical terms.
ivanbakel is not making some blanket claim that things always require complex technical terms. He is just saying that language -- including field-specific jargon -- evolves so that people steeped in the language can explain things to each other.
The clearest example of that how hard it is to decypher popular articles about quantum mechanics. Here, even when explaining concepts at a skin deep level, stories they tell are utterly inadequate if not misleading.
That's because even the basic concepts of QM are ones from linear algebra. And is shy Feynman was a sometimes devotee of the "shut up and calculate" school of interpretation.
Of course, but when explaining something to another person, one should strive to optimize for being understood, not for the highest information density.
If you skip out the part of the analogy where you actually return to the complex terminology you're simplifying, you haven't produced understanding - you've only made a just-so story which is a reflection of the reality, not a justification for it.
Great epistemological question for the ages. Vectors for the response include audience type, previous knowledge context, goal for how much needs to be conveyed...
The same thing that it means to hear a bad quality 8bit, 20kbps copy of a song. It's still better than NOT hearing it, and it gives you an idea about what it entails, even if you don't hear all the information in it.
Even in everyday language we make the distinction about understanding something fully and having a rough understanding of it.
> If you skip out the part of the analogy where you actually return to the complex terminology you're simplifying, you haven't produced understanding
Billions of people know what a telephone does and how to use one without understanding τῆλε ("tele-", distant) and φωνή ("phone", voice, talk, language). They all get the concept and nobody would argue that they need to understand the terminology as well.
Or: I can teach you to play chess without teaching a single chess-specific term. Yes, other players will laugh at you when you call the pawns "smurfs", but when you understand how to handle your smurf-army, opponents will stop laughing after hearing you announcing "death by smurf-supported horse in 2 turns". (Though they'll definitely hate you for calling a knight a horse ;).
But that doesn't remove the terminology, it only changes it. You can understand chess equally well in English, French, or Lebanese, because the syntax isn't as important as the semantics. You can't teach me chess without talking about the idea of a "piece", no matter what you end up calling it. There has to be some common groundwork.
>Billions of people know what a telephone does and how to use one without understanding τῆλε ("tele-", distant) and φωνή ("phone", voice, talk, language). They all get the concept and nobody would argue that they need to understand the terminology as well.
The etymology of words tends to have no bearing on understanding them technically. To me, that's a non-comparison.
What's really meant is more that plenty of people use a telephone without a comprehensive understanding of EM radiation. That doesn't mean they can't use their phones, it just means if you explain EM radiation to them in terms of phones, they probably aren't going to come away understanding it any better.
First it was that, to understand, we have to "actually return to the complex terminology you're simplifying". I disagree.
Now it is, that we need some common way of communicating. I agree.
But that are two different questions.
> [explaining EM radition in terms of phones does not help]
Nor does explaining EM radiation in terms of oranges. But I'd do neither. I'd first try to figure out whether you struggle with the concept of what a phone is or with the question how it works. Because step 1 in being understood is to understand what you are trying to say, to define the metric you can later use to evaluate your success in your attempt at being understood.
http://norvig.com/spell-correct.html
Norvig is a genius, and his genius is reflected in how he can so succinctly and clearly explain how to write a spell checker, without falling into elaborate jargon or droning on without getting to the point.
And that goes equally for his code and his prose, by the way.
There are many people who use a lot more jargon and verbosity to make their points, who are not in the same league as Norvig in terms of intelligence.
"The production volume of agriculturists is the reciprocal of the intellectual capacity of the producers."
also known as:
"The less intelligent the farmer, the bigger his potatoes."
That's what I hated the most with uni: Some professors could write 10 blackboards full of formulae, but could hardly put together a few words to explain what we were actually doing. Unsurprisingly, I don't remember much from these lectures. Fortunately that is not all professors, but it worries me how large a percent of them were like this.
The notion here is that educating beginners about a subject is different from communication between experts. Yes, experts use jargon because it is more efficient. But, the notion here is that if you truly understand something, you should be able to find a way to explain it to a non-expert, building up from words and concepts they already understand. And, further, if you can't do that, then perhaps your understanding is not as deep as you thought it was.
Was he referring to the Z state? If so, it's not really ternary logic but rather a "no result" state for cases where there is a control signal involved (e.g., in a tri-state buffer).
Even more non-binary things happen inside DRAM and flash memory.
DRAM is analog-based due to the fact that the core memory unit is a capacitor, so it can't be ternary (not digital). I might agree with flash though.
Algebra and calculus were one bleeding edge research that only the elite understood, now kids are learning it in middle school.
Yes, but knowing which information you can lose is why it is said that you don't understand it until you can explain it. Explaining it in simple terms forces you to over-simplify, and then you have to pick which information is important and which isn't...and doing that requires a deep understanding of the subject (and your audience).
It's about explaining in a way the recipient can understand so "simple terms" becomes "terms your recipient can understand".
And there are three kinds of explanation:
1. visual
2. mathematical
3. linguistic
So sometimes, you understand something visually, or mathematically, but you are forced to put it into verbal terms (say, over a text only channel, or voice), and then you may seem not to be able to explain it even though you understand it.
Next time I see this concept in action, maybe I'll suggest alternate methods of "explaining".
http://scottmccloud.com/4-inventions/triangle/index.html
http://homes.chass.utoronto.ca/~mfram/Media/0505-UC-triangle...
Filed under "I believe, but cannot prove": The three-way tension is a recurring pattern.
When trying to understand new things, I often try to reframe things as a triangle. When pondering intractable problems, I favor three-way solutions. Some quick examples from memory...
Project management: time, money, scope.
aka Quality: Fast, good, cheap.
US Govt balance of power: Executive, Congress, Judiciary
Language design: imparative, declarative, functional
Pop music: harmony, melody, lyrics
If you have to budget something, it becomes a soft trilemma: I have 24 hours for work + leisure + sleep, so any given combination will be a point inside a triangle whose vertices are (0, work), (0,leisure), (0, sleep). This structure is called a (2-)simplex.
BTW: a regular dilemma (should I spend or save) is a 1-simplex, and is a simple linear combination asave + (1-a)spend.
Now, this is fun for two reasons:
- You can have (n)-lemmas (i.e. n-fold tradeoff structures) that are modeled as (n-1)-simplices. Actually useful: you can do statistical inference on simplices using the Dirichlet distribution.
- A simplicial complex (basically, a set of simplices) can be used to build topological spaces part by part. This kind of maths (algebraic topology) is a whole "south part of the mountain" climb towards abstract mathematics that bypasses a lot of Cantorian handwringing on the ultra-local structure of topology. Instead, you're computing stuff from the get go -- and indeed one of the emerging machine learning techniques goes precisely from building a simplicial-like complex from data and computing characteristics of its topology.
And in my experience, the harder the subject, the more informally experts speak. Partly, I think, because they have less to prove, and partly because the harder the ideas you're talking about, the less you can afford to let language get in the way.
Informal language is the athletic clothing of ideas.
[0]: http://www.paulgraham.com/talk.html
I'm a physicist, and if you asked me to explain Newton's laws or classical thermodynamics to someone, I could probably give them a reasonably complete overview in a few hours. And in doing so I'd probably introduce and explain some jargon (entropy, derivatives, etc) along the way to make it easier.
On the otherhand, if you asked me to explain Quantum Field Theory or General Relativity in the same time, I'd basically just give up and throw out some friendly but inexact analogies about stretched rubber sheets and the like, and hope the walked away with some vague sense of the outlines of the subject.
I think PG has confused the informality people have when they have a common understanding with something else. Do you really think that a pair of experienced mechanics in an auto shop discussing a basic repair are going to be using more formal language? Turn one of the mechanics into an apprentice and consider the same discussion.
"Hell, if I could explain it to the average person, it wouldn't have been worth the Nobel prize." [1]
Showing a limitation of the maxim or Feynman's hubris?
[1] https://en.wikiquote.org/wiki/Richard_Feynman
But seriously, if you can't explain what a complex-valued "probability" is to a layman, apparently you don't understand quantum mechanics?
https://www.youtube.com/watch?v=AEUcmKDaklY
(Highly recommend the whole Feynman series, which is excerpted from an interview video called The Pleasure of Finding Things Out, which is also spectacular)
Obviously, if you use the "hairy" physics etc terms, then the layman wont understand your explanation, because he is not familiar with them.
I've seen people from my domain making ridiculous, inaccurate analogies and oversimplify to hide the inherent challenges that a task usually involves. And they get away with that because they aren't the ones actually building the product. They just have to sell it. And the one's they are selling to aren't/won't be using the product either, they will tell someone else to use it.
At least as good as math (and capturing the objective complexity better I think) is a working program. "What I can't program, I don't understand."
When people ask for a simple explanation, they usually expect it to be easy for them too, because we all want simple and easy at the same time, even though only one of those is objective. If your reply to "how do magnets work?" is to start by writing down Maxwell's equations, you're gonna get crap for it, but someone who uses the fake rubber band analogy will be well received. But who really understands magnetism better?
http://www.sciencealert.com/watch-richard-feynman-on-why-he-...
Trying to explain a concept in "simple terms" forces you to view it in terms of its connection to other, well-known phenomena.
At this point it's turtles all the way down.
That's what it looks like if you only look at the formula. (And why I'd say that formulas are not an "objectively simple way of expressing something*)
If you can somehow keep in mind what the formula is supposed to represent, the operations on characters will let you find some insight about that thing and will stop standing only for itself.
Also, mathematical formulas are essentially a different language. If people don't speak it, then it's a moot point.
You have to arbitrarily decide where to end: typical 5-year-old, typical high-school student, typical math professor.
What people mean by "simple terms" is "simple terms that I understand".
> Feynman was also quoted as saying:
> I think I can safely say that nobody understands quantum mechanics.
This, of course, does not hold for the inverse of 'what I can program, I do understand'.
The most useful meaning is "Have a consistent and reliable model for."
The code implements a model. Of course you can implement a very bad model as easily as a consistent and reliable model.
[1] https://en.m.wikipedia.org/wiki/Pushdown_automaton
As you say, informal explanations are great for understanding and bringing clarity, a lot of the time they're just used to help expand the unfamiliar math, but also a lot of the time they trade off precision, conciseness, or accuracy. At worst they complicate things with twists and relations in the informal explanation that don't exist in the real deal. Sometimes that might be helpful in the same way technically unrelated mnemonics are but it's important to point them out.
The conflation of simple, easy, concise, precise, clear, intuitive.. is the root of the issue when discussing what makes a good explanation.
This is also true for literature, poetry, non-analytical philosophy etc. And there the challenges are not about the complexity of the subject, but about its subtlety, and how one needs historical context and/or certain life experiences (or even proclivities) to be able to grasp a particular piece of work.
Taking mathematics as the operative example. A lot of deep theorems are trivial when you fully understand all relevant definitions. These proves cannot be stated 'in simple terms' because it took a stack of non-simple definitions to even formulate such a theorem.
Explaining the entirety of the stack gets hard when the stack gets deep. For example, a lot of math gets really hard to explain when the concept of the real numbers being 'uncountably infinite' whilst the rationals are 'countably infinite' isn't obvious.
A lot of deep theorems comes down to properly choosing the definitions so that the theorems are true.
[1] http://www.nobelprize.org/nobel_prizes/physics/laureates/196...
QED does explain quantum electrodynamics without any calculus or other math (for example, he uses imaginary clocks as a way to explain wave interference), but I think if the average person read it they wouldn't walk away with even a partial understanding of it.
She'll still have no idea what an interface actually is, but she'll have a good grasp of the concept.
Edit: nevermind. I was way off.
Ce qui se conçoit bien s'énonce clairement, et les mots pour le dire arrivent aisément.
https://www.youtube.com/watch?v=MO0r930Sn_8
Monkey eat => Monkey live.
Monkey live => Monkey eat.
Monkey not eat => Monkey not live.
Monkey not live => Monkey not eat.
Here "=>" is used as in "implies"/"because". The last statement is weird. There are more ways for monkey to "not live" than to "not eat".
Not being able to explain does not imply not being able to understand. Not understanding surely implies not being able to explain.
Correlation, Causation, get it ?
A statement like "if not a, then not b" is not applicable unless a "implies" b and a "is because of" b.
Correct me if I got it all wrong somehow.
Some people look at advanced mathematics or physics and wonder why it has to be so complicated and so full of jargon. It's complicated because it is. The jargon, believe it or not, is mostly an attempt to make it easier to communicate. It would be very, very difficult to wade through these ideas without introducing new words with precise definitions.
Then again, John von Neumann said, "In mathematics you don't understand things. You just get used to them." So maybe the title is true for trivial reasons after all.
He was not talking about writing a research paper on the simplest possible terms. Rather, about "explaining" a subject. When we are explaining, we always have to deliberately omit details to get the basic idea across, so that the other person can form the initial mental image. Sometimes we also have to resort to metaphors and comparisons from entirely different fields, or relate them to real world concepts.
Once that is done, then the lengthily process can begin on improving the fit of that mental image to the subject. This will require more explanations, more focused, in a different level of detail.
Case in point: dimensions higher than three. Most people cannot conceive what the heck a tesseract is, or why it is usually drawn the way it is. But then you can tell them to watch Feymann's Flatland scenario, and they will be able to understand the basic concept.
Are they mathematicians now? Hardly. But now they have some understanding on the subject which they did not have before, even if it is on a very high level. And can try to absorb more details, if so inclined.
I would tentatively agree that there is a skill which allows you to break down complex subjects into digestible chunks, but that usually depends a lot on the student as well as the teacher(e.g. some people are good at following metaphors, others learn very well from repeated examples, etc), but the outright statement in the topic suggests that there is no such thing as irreducible complexity, which seems ludicrous.
Nobody's having any delusions that explaining is the same as teaching. You can explain multidimensional analysis simply if you understand the subject fully, however you can't teach something to someone else unless they work with you and try it out themselves.
But if you want to explain to your politicians that hospitals need special machines to do radiation therapy you don't enroll them into nuclear physics 101, they can probably do with a very broad explanation of how radiation works.
Maybe explaining a complex concept in simple terms would take you 2 hours, 12 hours or 3 days, but you can do it if you really understand it.
We've started doing Explorable Explanations / Animated Explainers, here are some we've done and some that others have done:
- Explaining how GIT works: http://gun.js.org/explainers/school/class.html
- How neurons work: http://ncase.me/neurons/
- How end-to-end cryptography works: http://gun.js.org/explainers/data/security.html
- How gerrymandering works: http://polytrope.com/district/ (by a friend of mine!)
- How sorting on partial data / data streams works: http://gun.js.org/explainers/basketball/basketball.html
And more! It is possible, it can be done. But it is hard. That is no excuse for not trying though. Big shout out to Bret Victor's work for starting a lot of this, and thanks to Feynman for encouraging and practicing what he teaches.
Sadly it's 2017 and the popularity of TEDTalks make the laymen think otherwise.
The Feynman Lectures are now on Youtube[0], and I like to watch them (all of them) every few years. I highly recommend that if you've never seen them, you take some time and watch them- really watch them. Close the other windows, turn your phone to do not disturb, and really watch these masterpieces of education.
[0] https://www.youtube.com/watch?v=j3mhkYbznBk
You link to: https://en.wikipedia.org/wiki/The_Character_of_Physical_Law
"Feynman Lectures on Physics" are: https://en.wikipedia.org/wiki/The_Feynman_Lectures_on_Physic...
Are you serious?!?!
Oh shit, that's... I've been grossly misformed for a very long time.
Opposite example: Simplify how walking works, and make sure to include the critical systems such as major muscle groups, stabilizers, vision, inner ear, thigh/knee/pelvis/hip construction, the curved spine and its connection to the head, and blood pressure flow/regulation.
You'd be missing a lot of experimental data that has been collected for decades. You'd be missing the metallurgy, etc.
The basic principle is very simple. Maybe you could even build a low powered version, given materials and time.
If you insist that understand how to obtain the proper air to fuel ratio is required to understand how an internal combustion engine operates, then I'd say we don't really understand anything at all. Can you mine and refine the ore to create the allows it is made of? Can you program the ignition (or alternatively, build a carburator)?
We all need to stop at some point.
Take it a step further and the engine is built, but you need to make a modification (forced induction, race fuel, etc). If you don't want your engine to blow up, now you need to understand air/fuel mixture. Sure, I can do something with a simplified understanding. But the lack of a complete understanding will leave me with a blown-up engine.
Does it follow that if I understand metallurgy and internal combustion engines, but can't simplify it, that I don't understand it? Of course not; the entire premise was that I already understood it. Simplification does not show you understand something. It can only provide a shallow, incomplete understanding.
So you don't need to be able to simplify to understand. You need to understand to understand. You need to simplify to provide a simplistic, shallow view.
The question, of course, is how well this experiment has succeeded. My own point of view — which, however, does not seem to be shared by most of the people who worked with the students — is pessimistic. I don’t think I did very well by the students. When I look at the way the majority of the students handled the problems on the examinations, I think the system is a failure. Of course, my friends point out to me that there were one or two dozen students who — very surprisingly — understood almost everything in all of the lectures, and who were quite active in working with the material and worrying about the many points in an excited and interested way. These people have now, I believe, a first rate background in physics — and they are, after all, the ones I was trying to get at. But then, “The power of instruction is seldom of much efficacy except in those happy dispositions where it is almost superfluous. ” (Gibbon)
Richard P. Feynman, 1963
http://feynmanlectures.caltech.edu/I_91.html
Note that by his own account, most of his students did not do well. James Gleick's biography of Feynman, Genius, has a longer discussion of the disappointing results of his lectures to undergraduates at Caltech, many of whom reportedly stopped attending the lectures as they were not getting anything useful out of them.
That Feynman in fact had difficulty explaining freshman physics to the highly qualified students at Caltech surely does not indicate he did not understand freshman physics.
Some topics are simply very complex. It is not clear that they can always be conveyed in simple terms. In some cases, a "big picture" explanation may be possible but the details remain complicated. In some cases, a hand-waving analogy to some everyday phenomenon may create the illusion of understanding but be misleading or wrong.
To give a specific modern example, a state of the art video codec such as H.264 is extremely complex, built of many complicated components and sub-algorithms. While it may be possible to explain the big picture in relatively simple terms, the detailed implementation and operation is not simple. The inability of someone who creates or implements a video codec to explain it in simple terms to a layman is not an indication that they do not understand it.
He would be proving something in class, someone would stop to ask a question, and from their question he would know not to try to answer it as asked, but rather he knew they were asking the wrong question, because the question was predicated on a misunderstanding of something three or four steps back.
Now these were proofs that a working mathematician could probably do in their sleep, but they were a level up in terms of abstraction for most of the students and to be able to drag us along with him was a rare talent.
Even after I got my undergrad degree in physics I would reread the Lectures and would see things on a whole new level than I ever understood them the first time.
Edit: even "Surely you're Joking" had jokes that only someone that had graduate level general relativity would get (talking about "gee-mue-nue, gee-mue-nue")
We now have programs that "understand" certain domains better than humans. And yet there are no "simple terms" to explain their understanding, reduce it to a neat closed-form formula.
Isn't it a mistake to assume a nice, elegant explanation, a cute human "narrative", must always exist?
"But Radim, the computers don't really understand the domain!". Sorry, a cop out, No True Scotsman fallacy. In what way do they not?
This is crystal clear with Go: the little stories humans have built to deepen their understanding of the game over 2,000 years are irrelevant to a game's outcome. The details and variations are all that matters, and there is no way to reach the computer's level of understanding of the game without understanding the details.
Take for example legal concepts like securities law or environmental regulation. Yes, you can "simplify" an explanation of the Securities Act or the Paris Accord enough to fit them into a tweet, but you lose information necessary to formulating a full understanding.
If you're trying to have an informed debate about policy adoption, the details matter.
This incompatability has caused me problems my entire career. I've more or less stopped trying. I just do what I'm told and save my creative energy for my personal projects.
If you hold a complex idea in your head translating that into English can be difficult because part of the process is removing/altering information to fit into existing notions. That is why buzzwords are popular they can take an idea and put it in a relatable concepts for the masses.
This is the sort of thing you'd believe if you were an arrogant 20-something who thought they could learn any subject in a few hours, cushioned thoroughly by the illusion of understanding.
"Oh yeah, I understand the mechanisms of human vision. It's just rods and cones."
"I understand the causes of the American revolution. It was just people protecting their property."
"I understand Joyce's Ulysses. It's just follows three people from Dublin over a single day. I read the Cliffs notes."
"I understand why coffee makes me alert. It's just blocking some brain things that make you sleepy."
Now, I will agree that if you don't know how to break interactions down into teachable parts, you will probably have trouble as an engineer or scientist both advancing your own knowledge and introducing people to the field. But to suggest that your understanding of a subject hinges on being able to deliver an explanation in simple terms is just silly.
Complexity is real, but understanding means being able to conceptualize a system at such a high level that it becomes possible to convey the most important key ideas simply. This may involve some amount of over-simplification (which should be indicated if present) and will undoubtedly leave out a tremendous amount of detail, but the conceptual essence should be there.
The inability to do this is IMHO a sign of either lack of true and deep understanding or communication ability.
There are fields where nobody can really do this. These are fields that nobody truly understands yet.
Edit:
Let me give you an example. I had a bad (but typical) math prof in my very first calculus class who covered the board with examples and took us through how to differentiate a function. I was completely confused and stuck. Then I called my father on the phone and asked him and he told me "a derivative is a rate of change." Then I got the rest of calculus.
The prof had talked about slope but that wasn't conceptual enough. I needed "rate of change." That is the essence of calculus. It's the most important concept. Calculus deals with rates of change and the inverse (integrals). Once this concept was communicated the rest of calculus became possible to understand.
If you can't do that for a given area, you don't understand it or (as was the case with this math prof) are a poor communicator.
I think the disagreement hinges on whether you believe the snippets from the GP post e.g. the "rods and cones" or caffeine examples are fundamentally valuable or not.
> There are fields where nobody can really do this. These are fields that nobody truly understands yet.
Curious what you would consider an example of such a field?
FWIW my anecdotal example of what I think the GP is complaining about... I read "The Elegant Universe": https://en.wikipedia.org/wiki/The_Elegant_Universe
It's a good book, and I enjoyed reading it. But over subsequent years I've had growing doubts over the true value of the "understanding" that it provided.
The frontiers of physics always strike me this way. The names and explanations are incredibly strained and bizarre. This is probably the canonical example and is perhaps why The Elegant Universe doesn't quite achieve what it sets out to achieve.
Another one from my own academic past is "complexity theory" where hand-wavey concepts like "emergency" or "higher order causation" get thrown around with a lot of confusion. We can clearly see there's some "there" here, but I don't think we really understand what it all means yet. Put stuff together, interesting higher-order stuff happens, and so now let's create a combinatorial explosion of jargon around it. Publish or perish.
Finally "consciousness studies" or anything else semi-fringe probably counts. Again there's certainly some "there" there, but we don't even know the right questions to ask. Something is happening to matter that makes it wonder what's happening to it.
I suspect a so-called expert's ability to solve problems within a domain (i.e., how "expert" they are at that domain) would be a function of their ability to manipulate terms in that domain which to me is akin to their ability to those terms clearly.
I can't imagine someone easily solving problems within a domain wherein the terms they use are giant spaghetti balls of complexity. Anything can always be broken down into further levels of abstractions (or prove me wrong on that).
For every complex system, there is usually an underlying simpler, though not accurate, model that makes grasping the broader concepts easier.
So, if you think you understand something, it's worth trying to explain it simply, so you can push your own understanding of the subject.
One doesn't stop at that model. (Feynman didn't.)
It's used as scaffolding, later torn down, to organize thoughts. By continuously proceeding downwards, in a fractal nature, filling in gaps, tearing and replacing where needed, the larger structure always stands ready to contextualise.
This helps hold the audience's interest. It also lets the mind, by itself, fill in gaps, promoting retention as well as pedagogical efficiency.
But I also do research, and I absolutely do not expect deeply technical topics like advances in aerospace engineering or gene regulation or cosmology to have "simple explanations" that I can understand. I accept that I won't understand them without years of effort, and I won't denigrate experts in those fields because of my limitations.
I tell people to come up with elevator pitches for complex work all the time, but I don't evaluate the quality of a thinker by the quality of their elevator pitch.
There's a difference between explaining the entire topic and just giving you a very broad strokes approach to get you started learning the subject on your own. Nobody's having any delusions about summarizing a complex subject to a few sentences to substitute learning it the hard way, but if you can't give a very gentle explanation that can ease people in to the subject so they can get started by themselves, I would say that you don't fully understand the subject.
Also, I would add that this explanation is not universal at all. You have to be able to gauge your audience and adapt to what's most suitable.
I agree cs 101 is not the world. I also agree some concepts cannot be simplified and it shouldn't.
My point was what you hinted towards when you say "distilling things down". My gripe is that most people that have taught or guided me have had no inclination to do that. I would have had a much smoother learning curve for it.
I also feel if there is a concept that can be taught by one person to another, it can be slowly be presented in increasing order of complexity than be presented all at once. That feeling of slowly building or adding complexity is underestimated and hard.
I agree cs 101 is not the world. I also agree some concepts cannot be simplified and it shouldn't.
My point was what you hinted towards when you say "distilling things down". My gripe is that most people that have taught or guided me have had no inclination to do that. I would have had a much smoother learning curve for it.
I also feel if there is a concept that can be taught by one person to another, it can be slowly be presented in increasing order of complexity than be presented all at once. That feeling of slowly building or adding complexity is underestimated and hard.
I do agree it is arrogance.
As an example, when we teach people how to fly a wingsuit, we essentially teach wingsuiting (which is a massively complex pursuit) as a set of nested arrays of increasing complexity.
For example, for a first jump course, the goals are as follows: - Exit safely - Demonstrate ability to navigate in-flight - Deploy
These are the most foundational aspects of flying a nylon dress out of an airplane.
After a few jumps, we'll add complexity to each of those, so to shift one point, it starts looking like:
- Exit in an unstable manner and gain stability in less than three seconds
In order to do that, you'll need to understand what causes a wingsuit to be stable, why it gets unstable, what happens when it gets unstable, and how to correct it. Additionally, you'll understand why I want you to do it in less than three seconds. However, at first, you just need to get out of the fucking plane. If I try to tell you all this extra shit you won't remember the foundational thing I need you to remember. You'll probably get unstable, you'll probably figure it out kinda, and you'll probably at some point deploy a parachute.
Teaching's hard.
"The limits of one's language are the limits of one's world." It's controversial, but the concept being if you can't speak about it, you really can't know it fully. This gets very fuzzy when experiencing art or emotions, but there is a cold logic to it that I appreciate, too.
Although, i would agree that some experts do a bad job of explaining things. They're still experts. I think you'd agree that non-experts don't really have a chance at explaining things.
[1]https://www.youtube.com/watch?v=eLQ2atfqk2c
Maybe, maybe, some times I'll do something clever.
Some times someone will have an insight, intuition. Like a market analysis. Or frame some useful questions for our recommenders.
Otherwise, all of the complexity that I encounter day-to-day is because of people's cognitive limitations. We make it hard. Misunderstanding, miscommunication, separation anxiety, dogma, arrogance, cya, whatever.
Nobody cares about a genius who understands something until he can really interact about the subject.
Humans work with communication. When you learn about something, you read about it and you process it. To see if you understand it, you must communicate it back. It's not really possible to guess how you brain just learned about something unless you ask that brain to reformulate it.
It's fine that it takes 150-300 hours of back-and-forth communication to transfer the ideas behind how it actually works. That seems like a reasonable amount of time to communicate a complex idea to a novice.
There's also layers of understanding:
The LHC bangs two rocks together hard to see what happens when they break.
The LHC use magnets to accelerate protons fast enough they break apart because it recreates conditions similar to the early universe.
etc.
The acceleration is actually done by electric fields produced in RF cavities.
And do you really expect politics to raise funds for such a project if nobody can explain to them what they are doing?
That's okay: politicians don't actually need to know how it works, they need to know why they should build it -- which is a) something they have domain expertise in and b) a much easier question.
That's one option. You could also apply the knowledge and see if what you think you understand has resulted in a mental model from which you can make testable predictions that correspond to the real-world behavior of the modeled thing.
It's sort of like saying that a mathematician who doesn't know how to prove their results to one who only knows the axioms doesn't fully understand their results.
Everything can be talked about on different levels of abstraction, including ones which use metaphors or contain inaccuracies in order to create a jumping-off point for further learning.
If you understand something very well, you're able to use various levels of precision + abstraction in order to best communicate it to your audience. You can "play" with it, without getting confused or off-track. If you don't understand it well, this gets very hard.
I think this is true for teaching and mastery in general, and not just for teaching about science or complexity.
I do agree that poor communication skills are not a lack of understanding, however, not being able to explain something I thought I understood is a personal redflag.
from the zen of python:
If the implementation is hard to explain, it's a bad idea.
If the implementation is easy to explain, it may be a good idea.
An important distinction to make here is whether you're teaching someone to intuitively reason about something or to logically calculate it precisely.
I think as long as someone grasps the basics of high school math and has a decent working memory, they should be able to learn how to calculate anything precisely given that you break it down into simple steps. If you don't know how to do that, you have no clue what you're doing.
Then there's intuitive reasoning. I think for most things, if you're familiar with a topic, you should be able to teach it in a way that makes sense. Of course there's exceptions -- some people are incredibly brilliant, but lack social awareness. I think that's the exception rather than the average case.
But not all topics are so easily reduced. There's probably some exceptions. The strongest that comes to mind is quantum computing. As many times as you can explain things like superdense coding, there's still a sense of "magic". Where the results of the math seem unnatural. And you go back through each step and try to figure out where things went weird. But each step is a logical progression from the beginning.
On that note, Michael Nielson's "Quantum Computing for the Determined" is one of the most well-taught courses I've found, and does a very good job of stepping the viewer through a very complicated topic. Michael Nielson clearly knows what he's doing :)
http://michaelnielsen.org/blog/quantum-computing-for-the-det...
Well, yes, if the only goal is computation. It doesn't take many instructions at all to be Turing-complete, and then you can compute everything a Turing machine can, which is lots of things.
That said, being able to calculate something can be very meaningless. You can try to do AES[1] by hand, if you'd like; it would be frustrating and very time-consuming, and then at the end of it you'll have gained effectively no understanding of its design or structure.
Using AES as a case study as well, one could explain the basic idea of encryption to a layman without too many troubles, but I can't imagine trying to explain anything about its operation or design to a layperson in a way that would make any sense at all.
[1] https://en.wikipedia.org/wiki/Advanced_Encryption_Standard#D...
No one should devote their career to simply understanding a topic. If all you do is load someone else's work into your head, what value you are adding? A career is a mixture of learning as well as exploring and discovering new things. Explaining just the parts that are known is a much smaller order than transmitting the effort of your entire career into words.
Note also that Feynman isn't claiming you should be doing this all the time, but that you should be able to. I've been programming for about two decades and I've learned a ton (and yet still have even more left to learn). I'd like to believe that, yes, I could explain almost all of it to a beginning CS student or even a lay person.
They might not have the patience for me to build up all of the necessary structure from one simple piece at a time, but I think I could. After, that's how it got into my head in the first place. The parts that I couldn't do a good job decomposing and walking through are exactly the parts that I probably don't have a good handle on. (For me, networking and operating systems come to mind. I know some of the jargon, but I don't really know how it all works.)