Write a Thousand Roads to Rome

post by Screwtape · 2018-02-08T18:09:21.192Z · score: 149 (52 votes) · LW · GW · 13 comments

Contents

  First, a preamble describing how my brain is finicky.
  Next, a suggestion that most brains are finicky.
  Are we raising the sanity waterline, or are we building a sanity waterspout?
None
13 comments

Epistemological Status: Pretty sure I'm on to something here, also very sure I'm restating the obvious, utterly confident that restating the obvious is the point.

Sometimes a piece of writing gets two very different responses. Half the commenters say something like "this is really obvious and a waste of time to write" and half of them say something like "this is revolutionary and an amazing insight I never would have reached on my own, thank you!" I think these posts that get this reaction are vital, possibly even more useful in some ways than a post breaking new ground.

First, a preamble describing how my brain is finicky.

When I was first learning Calculus in college, I had a hard time conceptually understanding what a derivative was. I could do the operation most of the time, but didn't know what that operation represented or what real thing those symbols on the page described. Someone suggested it was the rate of change and the second derivative was the rate of change in the rate of change, but that didn't mean anything to me. Someone else suggested it was velocity and the second derivative was acceleration, but I was honestly not great at physics either. I think someone suggested something about stock prices or economics, but I didn't have enough econ knowledge at the time to remember what exactly they said.

Then my professor recognized one of the equations I'd been messing around with on some scrap paper as describing the attack bonus from a game, and pointed out that the derivative of that would be how much better the character would get each level and the second derivative would be useful if characters got stronger at different rates.

"Oh, like an onion knight from Final Fantasy?" I asked.

The professor just gave a confused expression, but the concept had already clicked for me and later that week I was using derivatives at my gaming sessions as well as suddenly doing much better in my physics classes. The moral of the story is that if you want me to understand something, the best way is usually to relate it to a roleplaying game. I'm not the only one who thinks like this- a close friend of mine had no idea why I was worried about intelligence explosions until I reminded him of the Intellect Potion exploit in Morrowind. Yes, this is a silly way for a brain to work, but I think we all agree brains don’t necessarily work in sensible ways.

Next, a suggestion that most brains are finicky.

SlateStarCodex’s What Universal Human Experiences Are You Missing Without Realizing It? is a gift that keeps on giving. The first example is that you can be completely unable to smell, and not realize this until asked to write about how a peach smells. What makes that relevant here is that they must have been asked about smells before and they offered up an answer copied from people around them, but the particular way that question got asked this time caught them short and they realized they couldn't smell.

The idea of learning styles is a fad that came and went in a flash, but the useful parts struck a chord with lots of people. My brother remembers things best if he listens to them and repeats them aloud. I remember things best if I learn them as part of a story and then write them down on an index card. My roommate remembers things best if they draw a picture or diagram of some kind. These kinds of mental differences are really common, and there is no single best format for information to be presented in.

A written article revealing revolutionary insights will never get absorbed by my brother until the audiobook comes out. If I want him to know a thing, then writing better articles with better insights isn’t actually what I need to do; I need to read the ones I have aloud.

Are we raising the sanity waterline, or are we building a sanity waterspout?

If a few people in your society are literate, then they can copy the holy books to pass down knowledge for future generations and maybe even do a little written debate with each other in the margins. If more people in your society are literate, then you can start using a printing press for newspapers and leaflets. If most people in your society are literate, you get wikipedia. Some skills have incredible compounding effects when more people in a society have them; pick your favourite three rationalist techniques, and imagine a society where those skills are as common as literacy is in America today. I suspect that an alternate universe where every adult knew that arguments aren't soldiers, that beliefs should pay rent, and that nobody is perfect but everything is commensurable would be doing really well compared to us, even if it cost them some new breakthroughs!

Actually, I don't think writing these restatements costs you that many new concepts. Writing out an explanation of an idea one already holds is easier for many people than coming up with a new idea, and I find the process of breaking down a notion to explain it to someone else often improves my own understanding of the original concept. Small changes or additions to an idea can result in something new; Nobody is Perfect, Everything is Commensurable is at least half derived from Money: the Unit of Caring, but the other half seems to have done good work by providing contrast. I've never had a math professor who had made an original discovery in mathmatics but many of them taught me useful things, and Khan Academy isn't going to be writing new proofs anytime soon but they're still doing valuable work by providing another way for people to learn the things that already are known. There isn't a clean tradeoff between restatements and new concepts because these use different skills.

I'm not suggesting that repeating things using different words is always better than describing a new insight; I'd rather have one Eliezer Yudkowsky working on the big problems than a hundred slightly more rational Screwtapes, because sometimes what matters is the highest intellectual peak your society can reach even if it's just one person. For a lot of problems today what matters is the peak and you want to boost one or two people to those lofty heights. (I think of these as "sanity waterspout" problems.) That said, if I want to learn multivariate calculus or basic game theory there will be at least a dozen different ways of learning it, from a video with pictures to a terse few pages of a textbook to some goofy edutainment videogame. This is a good thing, and I wish rationality was more like this. Think of your favourite technique: how many different formats can it be found in? Alternately, think of your favourite format for learning something: how many techniques and concepts can be found in it?

For every concept we want more people to understand, we should want it explained in more ways. If we want everyone to be a little more rational, then we need to put concepts into mediums that everyone can understand.

13 comments

Comments sorted by top scores.

comment by alkjash · 2018-02-08T18:41:01.514Z · score: 44 (14 votes) · LW(p) · GW(p)

I had a good laugh reading the negative reviews of the Sequences on Goodreads. For every ten silent aspiring rationalists loving it exactly the way it is, there's one vocal critic saying things like:

The book is very informative and I really liked some parts, but it was a very difficult read. Why? Because Yudkowsky simply LOVES beating a dead horse. There's just too much repetition and sloppy writing, although the ideas are great.
It's just... It ought to be more concise and clear.
This one was really hard to rate as my reactions to each post and sequence varied from "jeez, this just nailed it" to "meh, why are you explaining the same thing over and over again".

Eliezer knew exactly how important it is to explain the same thing using different words again and again.

comment by Ben Pace (Benito) · 2018-02-14T12:08:34.255Z · score: 9 (5 votes) · LW(p) · GW(p)

I think of it more as "There are 10 pre-requisites for understanding concept X. Most people have 6 or seven, and then I write a blog post for each of the 10. Most people, most of the time, feel like they're reading a thing they already know, yet I did have to write all 10 to be able to get everyone to take the step forward together."

comment by G Gordon Worley III (gworley) · 2018-02-08T19:20:48.536Z · score: 36 (10 votes) · LW(p) · GW(p)

Nice. I picked up something like this idea from Anna. Basically the idea is that the words that help one person are not the words that help another, and so if you come up with different words to describe the same thing you should share them in case they help someone else. Thus there is value in writing your own self-help, even if it only helps yourself, since usually it also helps others.

Put more poetically, it doesn't take great genius to help somebody; it just takes being enough like them and knowing something they don't that you can explain it to them in a way that they understand.

comment by Viliam · 2018-02-09T00:09:21.826Z · score: 21 (7 votes) · LW(p) · GW(p)

Now put this into contrast with how science is typically done. There is one original description. Everyone is supposed to reference it. And you don't get any points for writing the same thing using different words.

comment by Qiaochu_Yuan · 2018-02-09T05:13:46.633Z · score: 51 (15 votes) · LW(p) · GW(p)

Compare: Chris Olah and Shan Carter on Research Debt.

comment by TurnTrout · 2018-02-16T14:21:29.128Z · score: 4 (1 votes) · LW(p) · GW(p)

What an Omega-send. Added to my morning reading list.

comment by Luke A Somers · 2018-02-10T01:37:00.544Z · score: 6 (3 votes) · LW(p) · GW(p)

Mostly, yes. Feynman gets a lot of credit for making QED comprehensible, even though he didn't make it in the first place.

comment by Qiaochu_Yuan · 2018-02-08T18:14:30.857Z · score: 23 (6 votes) · LW(p) · GW(p)

Yes. In addition to this, even if a thing is really obvious to many or most people it still might not be common knowledge, and it's good to write a post that can become a standard reference that most people believe most people have seen, that people can link to, etc.

What is your relationship to your username?

comment by Screwtape · 2018-02-09T15:35:10.914Z · score: 8 (2 votes) · LW(p) · GW(p)

Agreed. I would also love an easy way to ask or answer "Huh, I'm having a hard time understanding this. Do you know of any explanations in X format?" The worst case when writing something that's already written (assuming you do it well) is that you waste a bit of time, but you can probably expect that your configuration of words will happen to be a better explanation for at least one person than what exists.

As for the username, I played the devil's advocate one too many times and someone used it as a nickname for me. I was already a fan of "Screwtape Proposes A Toast" by C.S. Lewis and it fit the pattern for names I respond to easily, so when I made an account on Less Wrong I picked that.

comment by ChristianKl · 2018-03-27T12:07:13.557Z · score: 5 (1 votes) · LW(p) · GW(p)
The idea of learning styles is a fad that came and went in a flash, but the useful parts struck a chord with lots of people. [ ... ]
A written article revealing revolutionary insights will never get absorbed by my brother until the audiobook comes out. If I want him to know a thing, then writing better articles with better insights isn’t actually what I need to do; I need to read the ones I have aloud.

One of the reasons it was a fad that went was that the "useful parts that struck a chord with lots of people" don't seem to done well in scientific studies and those studies that were done suggest that giving different people different teaching based on their learning style doesn't improve learning outcomes. If you think there's a way that learning styles are useful, how do you account for the academic results?

comment by Screwtape · 2018-03-27T21:23:10.913Z · score: 5 (2 votes) · LW(p) · GW(p)

Because frankly my opinion of the current American education system is that it is slightly worse than chance at teaching people things regardless of the pedagogical method; just because a blind man can't shoot straight doesn't mean there's anything wrong with the gun.

I admit I haven't read any scientific studies on learning styles and would cheerfully read any two that you wanted to point me at. It's possible the studies you're thinking of didn't involve schools or conventional teachers, and if that's true then I'm more interested in reading them. My experience with learning styles was mostly firsthand in school, and there the educational system made a few strange choices that (I think) blunted the usefulness of the approach.

To use my brother as an example: the school classified him as an auditory learner, but their auditory teaching methods involved having him give oral presentations in place of certain essay writing assignments. This was a small step forward since he could speak clearer and easier than he could write (though its impact on his grades was wiped out by the fact that drawing was hard for him and they had him do the visual learning assignments as well) but at no point did the school offer him an audiobook version of something instead of the paperback.

My prediction for the studies is that they had a specific list of learning styles each with a different specific study technique and/or that they asked each student to use each technique. This resulted respectively in some students not improving at all (because the technique that would help them wasn't on the specific list) and/or the best students doing worse (because they had to spend time using techniques that weren't suited to them) with the result that overall performance declined.

There are a lot of different ways to try and learn something. (Listen to it on loop, anki decks, write an essay about it, try to teach someone else, etc.) I think that humans naturally vary in which methods of learning are easiest and most effective for them, and a good idea is to try a lot of different ways of learning material, to keep track of which ways work best for you, and then to use those techniques to learn everything you want to learn going forward. The index cards that help me learn best aren't in any of the official Learning Styles technique lists the teachers had, but once I found them I started using them for everything while not expecting them to work especially well for the median student. That's the useful core at the heart of Learning Styles, even if they were never used to great effect.

Does that fit with the studies you've seen and your own understanding?

comment by Ben Pace (Benito) · 2018-02-15T10:53:39.159Z · score: 5 (1 votes) · LW(p) · GW(p)

I think this post points to an important problem and solution. I like the quirky concrete examples, and I feel this helped give me concrete ideas for changing how I write. While I think I'd made explicit the point of this post in the past, it still helped me, which is the very point of the post (+5 points for recursion). Anyhow, for these reasons I've curated this post.

comment by habryka (habryka4) · 2018-02-09T18:38:35.857Z · score: 5 (1 votes) · LW(p) · GW(p)

Promoted to the frontpage.