Learning takes a long time
I recently realized that I had greatly underestimated the inferential distance between most of my readers and myself. Thinking it over, I realize that the bulk of the difference comes from a difference in perspectives on how long it takes to learn substantive things.
People often tell me that they’re bad at math. I sometimes respond by saying that they didn’t spend enough time on it to know one way or the other. I averaged ~25+ hours a week thinking about math when I was 16 and 17, for a total of ~2,500+ hours. I needed to immerse myself in the math to become very good at it, in the same way that I would need to live in French speaking country to get very good at French. If my mathematical activity had been restricted exclusively to coursework, I never would have become a good mathematician.
Math grad students who want to learn algebraic geometry often spend spend two years going through Hartshorne’s dense and obscure textbook. it’s not uncommon for students to learn interesting applications only after having gone through it. I find this practice grotesque, and I don’t endorse it. I bring it up only to explain where I’m coming from. With the Hartshorne ritual as a standard practice, it’s felt to me like a very solid achievement to present substantive material that readers can understand after only ~10 hours of reading and reflecting deeply.
It was so salient to me that one can’t hope to become intellectually sophisticated without engaging in such activity on a regular basis that it didn’t occur to me that it might not be obvious everyone. I missed the fact that most of my readers aren’t in the habit of spending ~10 hours carefully reading a dense article and grappling with the ideas therein, so that even though I felt like I was making things accessible, I was still in the wrong ballpark altogether.
Thinking it over, I’m bemused by the irony of the situation. Even as I was exasperated by some readers’ apparent disinclination to read articles very carefully and think about them deeply, I was blind to the fact that I was failing because I hadn’t put thousands of hours into learning how to communicate to a general audience. Seeing how large my blindspot was made me realize “Oh… just as I had no idea how much time I need to put into developing my communication abilities to reach my readers, some of my readers who appeared to me to be trolling probably just had no way of knowing of how much time it takes to learn really deep things.”
The tens of thousands of hours that I put into developing intellectually didn’t feel like a slog – it was fascinating. It was the same for all of the deepest thinkers who I know. If you haven’t had this experience, and you’re open to it, you’re in for a wonderful treat.
Off-topic...
That’s the whole point… You can keep saying that “Scott is not bad at math, he just didn’t put in the required 10k hours” but if the difference between those good at something and the rest of us is that to them putting 10k hours into learning that something doesn’t “feel like a slog”.
The way that I was learning math was radically different from how my classmates were—the two things aren’t really in the same reference class. If he had been surrounded by people of similar intellectual curiosity, some of whom were interested in calculus, his subjective feelings around learning calculus may have been totally different.
Yes, it was the same for me, I saw others memorize, say, the trig identities without understanding them, even though we were in the same class, with the same teacher. In other classes I was the one resorting to memorization, because the overall picture didn’t make sense to me (art history, literature). I attribute this to my not having the aptitude to internalize these subjects, just like those artsy types had trouble internalizing math and physics.
Same here. I loved math, people around me hated math… but if I had to learn math the same way they did, I would probably hate it too. (I am speaking about elementary- and high-school math here.)
The sad part is, the way they learned math is the standard way of our educational system; my way was the exception. I learned some basic things from older relatives before elementary school, then learned other things from books. I played with math in my free time, I explored, under no pressure.
Their way of learning was probably more like: “Remember this, remember that, and if you fail to remember all these hundred unconnected facts at the exam next week you will be punished.”
And, like eli wrote, it is also a question of time. I started playing with math before I started elementary school, which is why I had enough time to play. Later, at university, when we learned some things that I didn’t play with before, I started hating those parts of math. Because I was forced to keep running, despite not understanding things thoroughly, and I hate doing that.
Later, it is similar at my job (programming). When I am allowed to work at my own pace, I first explore the problem, then write the program; I am productive and I feel good. Unfortunately, more often than not, the management has a completely different idea about how to run things: they focus on keeping the programmers running so they don’t waste time “obsessing over philosophical details that provide no value to the customer”, but of course later these details make the whole stuff fall apart, and then we waste time and energy fixing bugs that should not have appeared in the first place, and we usually even don’t have enough time to fix those bugs properly, so they keep returning, and that makes the customer angry. So at the end, everyone is unhappy, because we were not allowed to take time to do things properly. (Although I suspect that the managers make a conclusion that this all happened because we were not running fast enough.)
It is a truism that learning takes time. The issue is that we punish people for taking time to learn thoroughly. Our training/education regimes are largely designed to turn out large numbers of semi-numerate engineers in a short time rather than to produce any number of fluent scientists of any period of time.
I think that’s because, when looking at the aggregate of society, it’s more efficient to bring people up to the level of semi-proficiency than it is to bring them to the level of expertise. If you have 100,000 hours of training to allocate, you get more bang for your buck to train 50 people to 80% proficiency than it is to train 10 people to the level of an expert.
The flaw, of course, is that “training hours” isn’t a finite, discrete resource. Any individual can opt to spend additional time of their own accord if they are truly passionate. The problem is, at the points in our lives when we have the most free time to spend improving ourselves (read: high school), we also have the least idea of what the hell we want to do with it.
I don’t think it’s only a matter of training time. Having to learn for an exam requires you to learn the concepts in a few weeks instead of spending two years on it. Quite often people forget things after they wrote the exam.
Distributing the learning over a longer time frame allows for deeper integration.
Only when they’ve got some minimal level of guidance, such as our Best Textbooks thread and a dependency graph of subjects.
I agreed with the first two sentences, but you lost me on the last one. I may be misreading you, but if I’m not...
There seems to be a meme among certain groups of smart people (mainly physicists, mathematicians, and computer scientists) that engineers are ignorant. This is irritating to me, as a mechanical engineer who doesn’t seem myself as deficient compared against any of the mentioned groups. In fact, it seems to me that my education is broader and more complete than the mentioned groups.
I agree that society does not encourage learning things thoroughly, but engineering is not usually an example of that. There absolutely are a large number of engineers who don’t particularly care to understand things thoroughly, but I don’t think this is so different for people in other fields (especially CS).
You’re right that I shouldn’t discriminate against engineers ;-). I was more intending to make a distinction between, let’s say, professionally applicable Bachelors-level expertise and research-applicable postgraduate expertise. A BSc/BE can be used to train a civil engineer who builds bridges, but it won’t bring someone up to the level needed to do materials physics research.
I appreciate that clarification. There are a small number of people (some affiliated with LW) who consider calling someone an engineer an insult.
I agree.
I like these posts, thanks for writing them!
What makes you think that spending that time on thinking is untypical for LW?
The only difference is that different people didn’t spend their time developing intellectually the same way.
Let me approach this from another angle. I’ve been playing the flute since I was 5. Counting the several competitions and the 5 hours a day I used to practice. I’ve spent more than 10K hours. I probably passed that when I was 12 or 13. I’ve now gotten to the place where I can go for a few weeks at a time and then pick up the flute and be able to play it with aplomb and ease without degradation in skill. Learning does take a long time and even though music is/was still my life, it can be a slog and there isn’t really a faster way to do it especially on very complicated things (like being a novelist) because of the nature of forging the new pathways in our brains.
In my mind this is a great argument for broad-based classical education that touches on a variety of subjects and lets students then specialize in a field of study. It gets your brain ready for learning, gives you plenty of knowledge and helps you with your communication skills so that you can do what I do: explain really complicated things to a general audience. It also gives you the chance to speak knowledgeably with other knowledge brokers. In some ways, that is what our society has lost with excess specialization. Look at this site, how many people would benefit from a holistic education in philosophy but are unwilling to put in the hours and study it takes to learn about the conundrums of philosophy?
I think is a great argument for broad education.
It looks to me as if you mean underestimated. (If not, then you’ve underestimated the inferential distance between at least one reader and yourself this time.)
What you said, thanks.
Data point/encouragement: I’m getting a lot out of these, and I hope you keep writing them.
I’m one of those could-have-beens who dropped mathematics early on despite a strong interest and spent the next decade thinking he sucked at math before he rediscovered numerical proclivites in his early 20′s because FAI theory caused him to peek at Discrete Mathematics.
Thanks :-).
Anyone know/have advice on:
making the most of time spent on the deep thinking?
methods to find the best/most-diversely-explained sources of information?
One of my reference experiences in that domain is trying to teach the idea of recursion to a bioinformatics freshman. It was right before the exam, after one semester of Haskell.
At that moment I understood why the university teaches that semester of Haskell. The idea of recursion is important and doesn’t come naturally to people. It’s just worthwhile to spend a semester with a language that most of the students aren’t going to use afterwards because recursion is that important and that hard for students to learn. It’s simply takes a few months to get that idea into a brain.
Excellent post. The question is how to learn how to communicate to a general audience. I imagine it takes practise moreso than study, and that most books would only provide insights into one aspect of the problem while ignoring others. But some that come to mind are: Pinker’s Sense of Style, some of edge.org The Oxford Book of Modern Science Writing. It also might be useful to attempt to reverse-engineer the writing of people like Richard Feynmann or Oliver Sacks, or the speech of people like Bill Nye, Karl Kruszelnicki, etc. Also, the principles of science communication are probably fairly related to the principles of all nonfiction communication so one could also look further afield.
I’m a bit confused—are you referring to one of your LessWrong articles? Were you anticipating that readers would spend ~10 hours reading and thinking about it before commenting?
Not an upfront investment of ~10 hours, but my implicit model was that people would be inclined to spend that amount of time after having read a really substantive article. I may have spent ~1000 hours thinking about Scott Alexander’s article Generalizing From One Example: it’s very profound. When I comment on an article, I’ve almost always at least carefully read every word and made sure that I understand what the author was saying.
Consider the discussion around me saying that I now know how to feel universal love and compassion like Martin Luther King and Gandhi. In the past I would have found this very exasperating, feeling like “it’s really not that hard to understand what I meant, these people care so little about epistemic rationality that it’s not possible to have a productive conversation with them.”
What I came to realize is that the standard of careful reading that I had in mind (similar to the level of attention that I put into checking a mathematical proof) may be entirely alien to most of my readers, so that I may have been totally misreading them as not caring about epistemic rationality.
I think “careful reading” mistakes what that debate is about. Perceiving status claims in your writing is picking up on elements of your writing by carefully reading it.
If you want people to ignore that layer of meaning you are asking them to ignore information. You are asking them to read less carefully. The fact that you didn’t make a conscious choice to deliberately make a status claim doesn’t change this.
Apart from status if you assume that learning deep stuff takes a lot of time your claim that you got to world class in a skill in a short amount of time invites challenge. The most likely explanation is that you simply underrate what it takes to be world class in the skill. Could you explain how more careful reading would prevent that interpretation? I don’t see how spending more time with the article would change anything.
What I meant was that my focus was on the fact that I now perceive their dispositions to be learnable to a much greater extent then I had thought, not on my quality relative to other people. Note also that there are two different interpretations of “like” in in this context—qualitatively similar and quantitatively similar, and that my own focus was more on the former than on the latter.
To the extent that the information was not already in the original article, I would consider an appropriate response to be a request for clarification: “what exactly are you claiming?” rather than a challenge as to why I believe something.that I may or may not believe.
Robin’s article Againt Disclaimers is highly relevant here—efficiency of communication is greatly restricted by the need to clarify that you don’t mean things that you didn’t say before even having engaged with readers.
If it’s true that you don’t claimed that you reach the level of Martin Luther King then why did you object to James Miller’s
by saying:
If it just about the quality and not about the level you have at it, I don’t see misreading on James part.
“What exactly are you claiming?” gives the other person no information about where the disagreement might be and where things are unclear. A challenge that provides more details on the other hand allows you to understand what your writing communicated to the other person and use that understanding to clarify what you mean.
This is especially true in communication that isn’t live. In live communications it’s possible to ask a lot of questions for clarification.
It’s also not like we Yudkowsky hadn’t written “Politics is the Mindkiller” which clearly explains why using political examples for non-political issues is a bad idea if you care about epistemic rationality. There a conflict between claiming that you are much more advanced than your audience and making basic errors like that .
You also speak about wanting to appeal to a certain aura. That means your listeners standards for what qualifies of worthy of that aura matter. It’s not up to you to define that standard.
If I’m reading you correctly, you’re misreading me in the exact way that I was describing!
What I was objecting to there was the use of never: I would guess that > 0.1% of the population is capable of developing it. I didn’t intend to make a strong claim about my current level.
Political issues aren’t triggering for me personally. My error here is specific to my mediocre social skills. I don’t claim to be more advanced than my audience with respect to social skills – I’ve acknowledged that they’re a major weakness for me.
You seem to be thinking that I was saying that people should view me as having an aura. That’s not what I was saying. I was saying that the aura that people associate with MLK and Gandhi is distortionary, because people forget that they’re human like you and me.
Feeling compassion is something that I consider to be in the area of social skills. Doing it right means that your body does certain things. It relaxes. It likely releases oxytocin. It’s noticeable to other people.
It’s more than just an absence of anger at the other person. I do see how the strategy that you describes creates a state where you aren’t angry at the other person. I don’t see how it creates genuine compassion or love for the other person. Those words are more than just metaphors. They are embodied feelings.
Quite a lot of people believe that to be the case for them. On the other hand if you put them into charged political situations they don’t think as clearly as before anymore. What’s makes you believe that you are immune to that?
As far as I understand you neither interacted face to face with either of those people. I haven’t meet either of them either but I have meet people who do have skills in that area.
I have the experience of a person trying to mug me without me being angry about them or negatively triggered. In that extreme situation I worked well. At the same time I know that I can’t do certain things I have experienced other people doing.
If you as a math phd would tell me that Terence Tao is just a human and we shouldn’t put him on a pedastal that would be fine. I grant you that you know enough about math and how to be good at math to have a meaningful opinion on that issue. I don’t grant that for universal love to a mathematician for whom social skills are a weakness.
Yes, technically someone who’s word-class is still a human but they are world-class. They can do things that other people can’t.
In case you are interested in my math background, I do pass the test of having done mathematical proofs in calculus. On the other hand I don’t have advanced math abilities. Just because I have access to a heuristic, I won’t you use the hammer for every problem.
I agree with many of your points. Note that we’ve moved some distance afield from the question of whether my post and comments were initially misread.
A large part of my thinking here is that if something that I write seems obviously wrong, there’s probably been a miscommunication—if it were so obvious that a commenter could notice a major flaw in ~30 minutes when I’ve thought about it for hundreds of hours, I would have caught it already! :-)
I don’t think this is a good approach to take.
Consider that “if it were so obvious that an outside reviewer could notice a major bug in ~30 minutes when I’ve spent hundreds of hours writing this code, I would have caught it already” is widely held as false in programming.
See my comment here. I’ve vetted my ideas in the course of conversations with many good thinkers. By now you’ve seen enough instances in which I’ve appeared to be saying something different from what I was intending to communicate so that you should give substantial weight to that possibility when I say something that seems obviously wrong.
I have a lot more than 30 minutes of thinking about the term “unconditional love”.
Imagine a math freshman comes to you. He spend 100 hours thinking that he has found a way to prove P=NP. After all the famous mathematicians are also only human. Will it take you 30 minutes to find the flaw in his argument? Likely not because you spent a lot of time thinking about math and how to do mathematical proofs while the freshman hasn’t.
It’s also not only thinking time. You likely would be less good at math if you wouldn’t have learned from capable teachers about how math works.
What I meant was in part that what I appeared to be saying to you is not what I believe. There are semantic issued involved (what do the words “universal love and compassion mean?”). I was in fact talking specifically about being able to overcome knee jerk negative reactions to apparent hostility.
In general this is a useful move, looking for an idealized version of what is being said, abstracted from irrelevant details. What is relevant depends on context. For example, a mathematical statement with a minor error should probably be understood to refer to its corrected version, which may be easy to find based on the surrounding discussion.
In the Martin Luther King/Gandhi example, the problem is not the idea of ignoring some of the details as irrelevant for the idealized version of the intended claim (about Jonah’s emotional state), but lack of convincing justification for their irrelevance for a different claim (about Jonah’s status on LW), a signal that is proving hard to disclaim. Even then it seems clear that these details are irrelevant for the intended claim and should be discarded from its idealized reading.
Depending on your goal it makes sense to focus on different information while reading. If you read a mathematical proof, status is completely irrelevant. It’s rational to completely ignore that layer.
If you read an article on a platform like this it’s useful to understand what’s driven the author. Jonah thinks that because he spend his 10,000 hours on training epistemic rationality people should pay more attention to his writing. Paying more attention to his writing means treating him as higher status. The status is not irrelevant for a reader because it influences how the reader spends his time.
To me it’s not a big issue. I usually don’t parse for status when reading LW posts. At the same time it’s there. As far as I understand Jonah suggests that when people derivate from “mathematical style reasoning” that means they aren’t reading carefully. That it’s due to not having enough training in epistemic rationality.
I parse the post as saying: “Jonah’s advanced skills in reasoning allowed him in a short amount of time to learn MLK style compassion.” That would be evidence that suggests that Jonah is indeed having advanced skills in reasoning and thus deserving of a high status. High status that results in people spending more time reading and contemplating his posts.
I usually don’t focus on the status layer when I read LW either.
In this case Jonah not only intends to make a claim about his emotional state. He also makes a claim that MLK is just a human, which indicates that MLK abilities aren’t as impressive as people who hold him to be “more than a human” believe. Given that MLK is a political figure, that claim is even more problematic than it otherwise would be.
There’s also the claim that his emotional state is desirable and that it’s desirability means that other people should copy Jonah’s technique that gave Jonah that emotional state.
Careful reading means picking up those 3 claims in addition to the claim about Jonah’s emotional state. That’s very different then mathematical style reasoning. In a mathematical proof you only have to focus on explicitly made claims. If you just focus on Jonah’s emotional state and consider everything else insignificant I don’t think you are engaging with the substance of his post.
Of course that doesn’t means that it’s always important to engage with every claim that’s made. I usually don’t parse LW posts for status or focus on those claims.
I’m trying to clarify the issue by introducing the distinction between different implicit claims, so that relevance for each of these claims can be considered on its own. A reader may be interested in multiple claims, so that if a detail is relevant for one of them, it becomes relevant for the reader. But when it’s relevant for the reader, it may still be irrelevant for some of the claims. Talking about irrelevance for the reader collapses this distinction.
When Jonah is pointing to careful reading, that includes awareness of various claims that are being considered and relevance of presented details for each of them. Clarification may address uncertainty about these claims individually.