I've noticed a trend among the rationalist movement of favoring long and convoluted articles referencing other long and convoluted articles--the more inaccessible to the general public, the better.
I don't want to contend that there's anything inherently wrong with such articles, I contend precisely the opposite: there's nothing inherently wrong with short and direct articles.
One example of significant simplicity is Einstein's famous E=mc2 paper (Does the inertia of a body depend upon its energy-content?), which is merely three pages long.
Can anyone contend that Einstein's paper is either not significant or not straightforward?
It is also generally understood among writers that it's difficult to explain complex concepts in a simple way. And programmers do favor simpler code, and often transform complex code into simpler versions that achieve the same functionality in a process called code refactoring. Guess what... refactoring takes substantial effort.
The art of compressing complex ideas into succinct phrases is valued by the general population, and proof of that are quotes and memes.
“One should use common words to say uncommon things” ― Arthur Schopenhauer
There is power in simplicity.
One example of simple ideas with extreme potential is Karl Popper's notion of falsifiability: don't try to prove your beliefs, try to disprove them. That simple principle solves important problems in epistemology, such as the problem of induction and the problem of demarcation. And you don't need to understand all the philosophy behind this notion, only that many white swans don't prove the proposition that all swans are white, but a single black swan does disprove it. So it's more profitable to look for black swans.
And we can use simple concepts to defend the power of simplicity.
We can use falsifiability to explain that many simple ideas being unconsequential doesn't prove the claim that all simple ideas are inconsequential, but a single consequential idea that is simple does disprove it.
Therefore I've proved that simple notions can be important.
Jump in the discussion.
No email address required.
Notes -
More options
Context Copy link
Falsifiability is a good thing. Make your beliefs pay rent, eh? Or, for a less normative example, see Belief in Belief. Carl Sagan’s point is quite easily summarized: why should I care about the proverbial invisible, intangible, non-interactive dragon in my garage? Yet even he chose to frame it in a narrative.
Consider this line from another Eliezer post:
So why might Carl Sagan have chosen a parable, rather than a minimum-length sentence, to encode his observation? He compressed a heroic number of assumptions into simple language. Today I can think “oh, that’s an invisible dragon,” and get immediate intuition on the subject. Even better, I can say it out loud! If you’re familiar with the story, it’s a very efficient exchange. If not, I can give you a link, adding Carl’s encoding to your repertoire.
This is why the early rationalist community was…enthusiastic…about recommending the Sequences. Well, this plus all the social dynamics that apply to any clique. It is efficient compression. It provides a scaffolding on which others can work.
Scott, back in his Yvain days, wrote a response to that Eliezer comment.
I’m throwing in all these links not (just) because I expect it to annoy you, but because I find it amusing that you’ve done all this work to agree with some blog posts from 2008. Why? What’s your angle? What rent are you getting from “simple notions can be important?”
Frankly, it sounds like you’ve got another variable floating around. An extra concept, telling you that “simple notions can be important” is, in turn, important enough to evangelize. It’s certainly simple. But I think it’s also separate from most of the things you might normally use to decide whether one blog post is generally more deserving of accolades than another.
Simplicity has value. It may be sufficient…yet I can’t say, with confidence, that it necessary.
True to OP's point, your comment is unnecessarily long.
>encode his observation
>compressed a heroric number of assumptions
He's telling a story to get his point across.
That's it. That's all it took to make the same point, which is obvious. The midwittery here is incredible.
And your namecalling and antagonism is not acceptable. Don't do this.
More options
Context Copy link
More options
Context Copy link
Who cares if you find it amusing?
It has not been established that I did agree with that litany of articles you linked, you just stated that. And how is anyone supposed to refute your claim? Presumably they would need to spend around 8 hours to read all that information, and then refute it, which you know nobody is going to do.
So you intentionally raised the bar so high as to make your claim virtually irrefutable. Congratulations, you "win".
More options
Context Copy link
More options
Context Copy link
I generally like simplicity as well.
However, simple writings can be bad when trying to engage with others in a discussion. I've noticed that people tend to cut down on the simplicity of other's ideas first, and their ideas second.
An exercise: Imagine I summed up your post by saying "simplicity is good", and then argued against that position by saying "some things are just complex and you can't get around it". You would probably be rightly frustrated because you'd feel that you addressed that point, but my summary simplified your explanation away.
Another exercise: anytime you think a post is too convoluted, simplify it in your head, then see if you can imagine someone having objections to the simplified version that are answered in the more convoluted version. (I notice this with Scott Alexander's writings pretty often, where I think 'I don't need all this extra stuff', but then see comments from people that didn't closely read the piece. They object in a way that was answered by the thing I thought was unnecessary).
Great point, but he actually has something worthwhile to say while the self-important bagpipes do not.
More options
Context Copy link
Because it's too simple. But if you try to do it in say two paragraphs you might be able to extract the gist of it.
I'm pretty sure I can come up with better versions of at least some of Scott Alexander's writings that are in fact simpler. I wouldn't be making the same points as him though.
People have too much ego though and think that their ideas cannot be explained better by other people, or even find it offensive for example if I claim I can explain something better than Scott Alexander. Why?
In open source projects programmers have to get rid of that ego, and other people constantly suggest ways to simply the code, sometimes rewrite it completely, and guess what the original author says... Thanks. I've made better versions of some big wig programmers and nobody finds it impossible or offensive. We all think differently and some people think of thinks we just don't. Why would that hurt anybody's ego?
Which is exactly what I said.
Then why didn't you do it! This is on a post where you are arguing about the benefits of simplicity and you can't be bothered to follow your own advice?
If you aren't making the same points as him, how is it a better version? It is a different piece of writing. That is like saying "I could write a simpler word processing program than microsoft word" and then you show us notepad. Of course it is simpler, but it is not the same thing.
Because sometimes you are removing features from a complex program, and claiming that you have made it simpler. Other people get annoyed because you have removed features.
I have about a decade of coding experience. This is not a new concept to me. But sometimes a junior programmer goes in a removes a critical piece of code to the functioning of the program, because they didn't understand why it was necessary. I have done this, and people under me have done this. Sometimes during code reviews I even thought "oh hey that looks much simpler than what I wrote, good on them". Only to come back and redo my complex set of code a week later when I realized what bug they caused.
In programming there is often a logical set of reasons why a particular piece of code exists. In writing that connection is a little more tenuous. Or if you want to maintain the coding metaphor think of a piece of writing as a bit like a piece of code, but it is a set of instructions for the human brain rather than a computer. Human brains vary quite a bit, and they are also a bit more emotional than most machines. So two obvious ways that writing would differ from code:
There would be parts of the writing that is necessary for some brains, and unnecessary for other brains. If a piece of writing is 'unnecessary' for your particular brain, it doesn't mean it is unnecessary for all brains. You could remove it, but you don't know who you have lost by doing so. Its possible you lost no one, its also possible you lost everyone else.
Emotions in the brain can change how people interpret things. Sometimes you need to set an emotional ambiance in writing to be interpreted correctly. To compare to programing, you need to import some packages first. But the brain is a run-time language, not compile time. So its just gonna take whatever whacky shit you give it and run with that. This changes writing by requiring introductory sections that set the mood. Scott does this in his writing, where he has an introductory section on controversial topics. Those intro sections are often meant to pull your head into the clouds, think big picture, and hopefully calm you down a little if you were coming in angry.
Because I cannot think what I do not think.
I did it as simple as I could. That doesn't mean other people cannot take my output and make it even simpler.
I do not have the mind of other people. Only my mind.
Because I would be making a more general point, that includes his point.
You don't understand what refactoring is. The codes has exactly the same functionality. That means no features are removed.
It does exactly the same thing, just with simpler code.
I'm not a junior programmer. I can spot when a piece of code is truly not doing anything.
Here's an example where I found a line of code that wasn't doing absolutely anything in Linux: lib/kstrtox.c: remove redundant cleanup. For some reason the best programmers in the word didn't see this in the core of Linux, but they agreed my assessment was correct.
I can show you much more complex examples where I reorganized the code and get rid of 50% of the code, and it still does exactly the same.
Why didn't they spot these issues? Because they don't have my mind.
Source code is for humans to read. Machine code is for computers. When I refactor code it's for other humans to be able to make sense of it in an easier way. When the source code is simpler, humans have an easier time understanding it and spotting problems.
I'm aware of how refactoring is supposed to work. Not everything goes perfectly with human endeavors.
I'm pointing out common failure modes, and your response is that they should just not fail.
And that's how it works, with people who do know how to refactor (e.g. I).
It's precisely the exact opposite: I'm saying it can work, especially by people who know how to do it. You responses imply that it's impossible to do it.
I said that people somehow find it offensive if somebody claims they can write a simpler version of something their idol Scott Alexander wrote and not lose anything in the process (like proper refactoring), and you seem to be offended by that notion. I did not even make the claim, I merely pointed out what would happen if I were to make the claim.
Is such notion so offensive that it must necessarily fail?
There is a failure mode that happens for code refactoring. I hope you agree that some people out there, maybe not you, can fail at things like refactoring.
Refactoring is often easy for code related things, because in the end there is a simple test for whether a refactoring has gone well: just run the code again and see if the results are the same.
Imagine if you were coding, but everyone had a different OS, a different browser, and different versions of the code libraries you were using. You can be certain that your code refactoring is safe for your machine at this given moment. But you don't know with certainty that it is also going to run on everyone else's machine.
My point has always been that writing is not so straightforward, and it is closer to the hellish existence where everyone has their own unique OS, browser, and code libraries.
The whole point of refactoring code is that it is doing the same thing in the end. My whole point about writing is that we rarely understand what the hell it is doing in the first place, much less what happens when we change it. Why did reading Scott's stuff convince me so easily, but when I shared it with friends it didn't change their minds at all? Why do you find Scott's writings too long, while I find the length just fine? It is because our minds are different.
I am not trying to argue the impossible case that no writing can ever be simplified. I would just say that the simplification is a lossy and imperfect form of data compression. And that what is being lost is probably not apparent to you, because you are probably stripping the things that your mind didn't need in the first place, but that others might have needed. If I or others seem "offended" that you claim to be able to write lossless compression of data, then think of it as the same "offense" that physicists feel towards people that claim to have invented perpetual motion machines.
Yes I do, because I follow good programming practices.
I can give you examples where I refactored code and I added unit tests to make sure that any and all changes I did retained exactly the same functionality the original code had. If it worked in someone's machine before, it should work in that machine afterwards.
This is not theoretical, I've done these refactoring, and the result works in millions of machines just fine. I can show you the commits.
Only if you don't follow good programming practices.
If you follow simple logically-independent steps, the process cannot fail.
But you can make a guess, and that guess can be right. That's what writing is.
No, not necessarily. Maybe 99.9% of the writers would lose something important in the simplification most of the time, but not all.
So you accept you consider it impossible.
You are not getting me with the whole programming metaphor. If you'd stop thinking that I was questioning your chops as a programmer for two seconds you'd maybe understand.
You have not committed a single piece of code in the theoretical world I made up where everyone has a unique OS. You have committed code in our world that just has basically three OS's you need to worry about.
In both worlds its not just about you following good coding practices or good writing practices. Its about the people writing the OSes also following good practices. In the theoretical world where there are a billion different OSes and they are nearly all written by amateurs, it doesn't matter how careful you are with your code, because its gonna be run on top of someone else's shitty code.
Does your program still work if the CPU doesn't know how to add 1+1? Or if the library running your code just randomly decides its gonna do garbage collection in its own special snowflake way and deletes a bunch of variables you need?
Your current programming ability relies on the fact that the computers it runs on are relatively stable and consistent.
Humans are not stable and consistent. Thus writing for them is not going to be the same as writing for a computer.
I shouldn't have written this last message. I'm done with this conversation. If you were correct about simple writing being effective then one of us should have convinced the other person in the first one or two exchanges.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Communicating with high information density is a hard-won art. I've spent years curating it for myself in the technical realm and coaching others to improve their own ability.
Many rat text walls contain too much fluff and strive for length. Some of this is correct and by design as others have mentioned. We don't need to limit ourselves to pithy 140-character hot takes, and most complex topics deserve more nuanced or detailed analysis.
One element you're missing here, though, is the value of flavor. We're here, at least partly, to be entertained (not just to communicate). Posts that include an element of ironic humor, to me, are the best ones - even if they're not strictly as concise as they can be.
Side note: I found myself trying to cut down this comment as much as possible, and have watched employees shift gears into speaking more slowly and succinctly during reviews where we've given them feedback on communication. Just an interesting phenomenon.
Except jokes contain information too. When two seemingly unrelated ideas are connected by the author, we find that funny. Jokes make ideas more accessible, and also more memorable, which makes them more likely to be remembered, and shared. That's why many quotes are funny, and so are memes.
More options
Context Copy link
More options
Context Copy link
Two competing Pre-Reqs on this: Watch This TikTok video and read This Borges Short Story. Both are short, the TikTok is a minute and the short story is 4 pages, so plenty concise.
Now compare the two. The notch on the book is so precise a metaphor that while the whole book is communicated by that notch, it is a code so complex that it would require absurdly accurate skill in measurement and mathematics to interpret. Even within the thought experiment, most people would assume that it would require some kind of complex microscope and computerized calculator to achieve. But let's go a step further. Let's imagine a superhuman who can see the notch on the metal rod and instantly measure the ratio visually, calculate the ratio, and interpret that into the written work. So that merely by showing such a man astick, I've sent him Funes the Memorius as surely as emailing you the pdf. But that requires both an immense amount of baseline skill, and a huge degree of education and recall. In understanding the code, and in the mathematics necessary to interpret it.
On the reverse of the coin is Funes, who Borges' narrator describes as
Part of what makes Funes the way he is, is that before his superhero origin story he is an uneducated farmhand. He is raw intellect undiluted. He does not have the mental processes to encode things in a quick and useful way.
The reason so many Rationalist/Adjacent writers, and Mottizens are in this group, write at such length is because they and their audience lack metaphors and symbols for the concepts they are discussing. Two well read and well trained philosophers or theologians can argue almost entirely in metaphors; Baridan's Ass or Schrodinger's Cat or You Can Never Step in the Same River Twice or the Melian Dialogue or Pascal's Gamble or a Faustian Bargain or Funes the Memorius.* A few words and a whole mass of concepts and ideas and metaphors floods the brain. To a well-read Westerner, just by saying Funes the Memorius (or in some cases, "hey remember that Borges short story you read in undergrad with the guy who remembered everything?") I have achieved the same thing the stick did in the thought experiment. But both require common education and skill. The rationalist writer is often stepping out of his own education, and his audience is a generally intelligent but not particularly specialized group of readers. Everything has be explained from first principles, like Funes' useless numbering system.
*This is why I'm in favor of a classics based education. I gives you common ground to discuss. And the metaphors you get from the Bible and the Iliad are critical to interpreting Faust and Dante, which are critical to interpreting all the literature that comes after that.
But we use those and more all the time. Pascal’s gambit and Schrodinger’s cat are almost too cliché for this venue. We have Pascal’s mugging, Moloch, a dane you can’t get rid of, a nazi who knocks, a cathedral, russell’s teapot, theseus’s ship, falsification, steelmanning, nash equilibrium, hypoagency, p hacking, oneboxing, backpacker harvesting, utility monsters, specks of dust, beliefs paying rent, pareto optimums, spherical cows, and, last but not least, the motte.
People’s complaints about this place are usually the opposite: that we use too much jargon, that we are being deliberately obscure. Or they will say the specifically rationalist concepts are not original. Which has a degree of truth, but if the point is to write more concisely, we still need to use the original term for the concept, and nothing will have been gained by switching names.
More options
Context Copy link
More options
Context Copy link
Einstein's paper is short because a longer paper is not needed. Something more complicated like General Relativity , QFT, or String Theory needs much greater exposition.
As Einstein also said, "Everything should be made as simple as possible, but not simpler"
Some posts come off as long-winded because of needing a lot of examples to explain a concept, or overquoting. I think instead of quoting from an entre page of a book, just link to it.
Unfortunately most links are clicked by a very small proportion of readers (e.g. I made a reddit post showing off this github page. The post got free 120 upvotes iirc I got 20 views on the github page).
More options
Context Copy link
Yes, but I'm not complaining about long articles. If an idea requires a long explanation, so be it.
I'm talking about short articles. The assumption around here seems to be that if an article is short, it must be because the ideas are not novel and/or trivial.
But that's not true. Einstein's idea that
E=mc2is novel, consequential, and simple.I think longwinded posts do well because people perceive that a lot of effort went into it. But this does not mean short posts are automatically dismissed.
Yes, but as others have accepted in this thread: short articles can take more effort than a long one. Writing concisely and succinctly takes more effort than just typing whatever comes to mind.
Doesn't it? Do you have an example of a short article that did well?
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
You missed an opportunity to say something like 'simplicity is good, stop writing long essays people skim through and misunderstand!' and just leave it at that...
That would be simple, but not what I contend.
More options
Context Copy link
More options
Context Copy link
I think pretty much everyone agrees that short + concise language is better, but it also takes quite a lot more talent to write that way. My read of the whole rationalist "wall of text" habit is that we are simply writing about extremely complicated things without commensurate talent. There are certainly a few people with both great writing skills and great technical skills, but if you want just the former, there are millions of books to read, and if you want just the latter then rationalist groups seem like an OK place to find it.
It depends. Moldbug is the opposite of concise yet is still talented.
I mean, I don't see how that contradicts what I said. You can be talented and put your efforts into other good qualities (such as prose, persuasiveness, or sensationalism) or just have a different kind of talent than the kind that's necessary to write concisely.
More options
Context Copy link
More options
Context Copy link
Agreed. But I think this tendency leads to a fallacious bias
wall-of-text ⇒ interesting,¬wall-of-text ⇒ ¬interesting.Have you see a simple article upvoted by mottizens?
I like walls of text so long as the information density is good. A short post, even with high information density, usually still doesn't have the time to explore its ideas fully. I've seen some simple articles upvoted by mottizens, yeah, but the long effortposts will always get more upvotes because they are genuinely better. This is because those who can write concisely often pack more thoughts into their posts rather than shortening them.
You are making a fallacy here. You are saying that
B (100 * 0.1)is not necessarily better thanA (1000 * 0.025)because even if the information density is higher inB, the total ideas are not as many asA.But you are assuming that
C (500 * 0.05)doesn't exist.You are also assuming an article needs a lot of information. But why? What's wrong with an article exploring a single idea, but an idea that is very important?
You also don't seem to understand what simplifying means. It doesn't mean compacting information, it means getting rid of unnecessary information.
For example recently I was discussing that a person claiming it's not profitable for them to lose $100k for an episode lost, while at the same time claiming the contract could be negotiated from $50m to $65m. The ratio is 150 mores of what they claim they "couldn't afford". I initially used an example of a $10 contract, but then I realized I can use percentages. They claim they cannot afford to lose 0.2%, but they can increase the contract 30%.
Using percentages the idea I wanted to convey it's still there, but a ton of unnecessary information is now gone.
Using bigger information chunks (one percentage instead of two numbers) the idea is easier to transmit. However, the information density is reduced, but not because it's "compacted".
Are they?
Will a long post always be better than a shorter post?
Nah, this isn't what I was talking about at all. 100 * 0.1 and 1000 * 0.025 both sound pretty bad to me. What I'm saying is that if you can write with an information density of 0.1 (which I assume is fairly high) then I'd prefer you keep writing until your post is at least bite-sized if not meal-sized.
There are diminishing returns to information condensation as well as to post length. Maybe a 1000-word post is 1.5x as interesting as a 500-word post, all else being equal. And a post with 10 information density is triple as hard as a post with 5 information density to write. Not to get too autistic here, but I think we can roughly model a given post's quality as
Q = density^1.2 * length
and the effort it takes to write it as
E = density ^ 3 + length ^ 1.5
So after a certain point you should stop optimizing for density and start lengthening your post instead, provided there are still things to say. There's a reason novels are so much more popular than poetry, despite the latter arguably having far greater information density.
If we follow this, then a 500-word post with a density of 10 is better than a 1000-word post with a density of 5. Even a post with 100 words and a density of 40 is better.
Of course you would prefer a 1000-word post with high density over a 100-word post with high density, but you as a reader don't get to choose. People write what they write.
What I'm saying is that if you see a post with 100 words, it shouldn't be discriminated in favor of posts with 1000 words, because clearly it's possible for the short one to be better.
This makes no sense. The effort should be proportional to both density and length, 500 words more at density 40 takes more effort than at density 5.
Moreover, if you start from this end, then you are going to be clearly biased, because you are going to assume that it's unlikely somebody spent 8 times the energy to achieve 4 times the density of a short post. You are just going to presuppose it's low quality from the start.
Arguing about the exact mathematics of the off-the-cuff equation is an enormous waste of time. Point is there are diminishing returns to density relative to effort after a certain point.
You're right, I should have multiplied.
OK, and? It's not like bias matters much here, so long as I still give a short post a chance. What are we even arguing about at this point? Do you substantively disagree that density has diminishing returns?
That's the only time when bias matters.
You are pretty much saying the bias of a jury doesn't matter as long as they give the defendant a chance. But it's a biased chance! Their verdict is likely to be wrong. If you are biased it's likely better to not judge at all.
Yes. It's completely up to the writer how much effort to put into an article, and a short article shouldn't be assumed to be low effort.
You tried to bring up math, not me, you can't say this is the math of how I think, and then say, "OK, that's completely wrong, but my point still stands".
If we follow your math, then you think it takes 16 times more energy for me to write a great 100-word text, than it takes to write an average 1000-word text. I don't think that's how it works.
The math doesn't work. According to your math it takes 32 times more effort to write a 1000-word text than a 100-word text at the same density:
(1000/100)^1.5. That's obviously not true, I think it's linearly proportional, so it's justx^1.Then, according to you, something with a density of 40 takes 512 times more effort to write than something with a density of 5:
(40/5)^3. Again, I don't think that's true. Once again I think it's linearly proportional.So the right equation is
e = density * length. In that case instead of taking 16 times, it takes 0.8 times:(40 * 100) / (5 * 1000). So it takes more effort to write the long average article. Then of course we would need to adjust the equation for quality as well.The math is so completely off that your point does not stand. You should not assume assume a short article is necessarily bad quality just because it would have taken a "prohibitively" large amount of effort to be of good quality.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
"I apologize for such a long letter - I didn't have time to write a short one." ― Mark Twain
More options
Context Copy link
More options
Context Copy link