YouSaid · the spoken record

Andy Matuschak

lines on the record: 158
first: 2023-07-12
most recent: 2023-07-12
sittings or episodes: 1
sources: podcast

Every line below is reproduced as it was said and linked to the record it came from. Nothing here is summarised or generated. Directory · Search · Corrections

“Unless they're public individuals. And so there's a lot of notes that are not public, but they're mostly not durable. They wouldn't be all that meaningful to others. The journals might be, but they're also intimate.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“I don't know by word count, by note. So like, for instance, my journal notes are not published, and there's one of those every day. So there's a lot of them. So if we're looking by note, we're excluding all of those. I also have like a note about all of the people in my life. And those are for the most part not public.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, you're right. So he's leaning more on his past explanations, which is interesting because he can't update them. Like that format of writing a newsletter and then linking to past newsletters, or as you say, this is sort of a former blogosphere thing to do. You have a series of six words and each word is a link to a previous post on the topic. I certainly have written stuff like that. It's kind of funny. I mean, it's approximating the durable note thing I was writing about, but without the ability to revise it over time. And maybe for many topics, like you don't need that ability, certainly the case that, well, I wonder now what fraction of my notes are basically in the state they were when I did my first major revision of them, it's probably at least a third. It might be more than half.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Often I'm recapitulating stuff I've written before, and I have these other things that are trying to. More durable, be like a useful reference that can stand outside of time. The combination feels useful. I don't yet have like a clear model of when one is better than the other.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Quite accumulates for either of them, at least not in the same way It's very interesting. They're doing the whole thing over again every day. One thing I find kind of interesting about Nat Levine's newsletter is that when he's talking about a topic repeatedly, like something that comes up, like the recent bank collapse or something, he will have to explain some concept like interest rate risk over and over and over again for days. Like every day he has to explain it. But every day he explains it anew. And every day the explanation is colored a little bit by that day. And this is an argument against the kind of note-taking that I do. It's an argument for ephemerality, for like recreating the thing every day because it will change and it will become kind of inflected by what you're thinking about now and your experiences. I think it's pretty interesting. I find myself these days doing the kind of mix. I have a journal that's about today and I'll do a bunch of writing.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“And it means that when I go to design the next system and I'm thinking about the role of questions in text, I'll have sort of a place to look. Role of hypertext in this is roughly as a kind of navigational aid as possible to do this without hypertext. You just end up with like what Lumann had a giant, something like a dresser, but made of card files rather than drawers for clothing.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Thanks. It does an interesting thing for me as a writer. It lets me build stuff up over time. So today I was working on, I was reading this very old cognitive psychology paper on the topic of adjunct questions, which we discussed earlier, the effects of asking questions while you read, not on remembering the information covered in the questions, but actually just on kind of the general effect that it has on stuff that isn't touched by the questions. I have some notes on the design decisions of the mnemonic medium, this quantum country thing that I was talking about earlier, interleaving the questions into the text. And those notes are kind of partial. They evolve over time. What was the impact of doing this? My notes about that, they've come from interviews with readers. They expand when I read a paper like this that's relevant to them.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“So there were all these dreams of hypertext novels, for instance, and some people wrote them. And one of the problems that a hypertext novel has is actually could be seen in a choose your own adventure book that existed before there was digital hypertext. And that's that the author is forced to write something like a lowest common denominator story, the page that is the destination of a hyperlink, it has to work as the endpoint of all of its reference. And so it can't establish any kind of coherent or consistent arc unless there's a kind of sameness all of the reference. And the more that there's sameness to the reference, like the less useful hypertext is. So a lot of people have been disappointed by this conclusion, I among them. I'll say that I do find hypertext very useful in my own notes, not really for reading.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Of chemical compounds and their properties. Hypertext is going to let you navigate that book very effectively. Likewise, dictionaries have been revolutionized by hypertext. So navigating around thesauruses by clicking on links to say like, oh, shade it a little bit more like that. It's like a much better thesaurus. So I guess I'm making the case that there are certain kinds of text that are more amenable to hypertext because they are more amenable to having the reader dropped in the middle of them. Encyclopedia are like that. Dictionaries are like that. Most text is not like that. And I guess like most concepts are not like that. I guess most ideas are embedded in something kind of holistic or richer. They require a narrative arc. They're difficult to excerpt. Not everything, but things that are not so raw and atomically informational.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“But you were expected to get the books down and look up the appropriate pages. This Entopicon wasn't that successful. I think it's in part because those concepts, unlike the Wikipedia entries, they don't quite stand on their own so cleanly. You kind of need sinews, you need linkages. And actually, I want to make the case that while Wikipedia is an astounding resource, I find it rarely to be the best available introduction or explanation of a topic. I find it often to be like a good jumping off point. It'll help me know the right thing to ask about. It's good as a reference. So hypertext is a very effective navigational aid. It can help you get to a spot that you're looking for very quickly because it's about automating, flipping through pages. And so for a reference, it's very effective. If what you have is a table of chemical compounds and their properties, like a book.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“This is a really good question. So, I think the reason why Wikipedia works as well as it does is that encyclopedia entries are already forced to stand on their own. And that was true before hypertext existed. In fact, encyclopedias were already hypertext-ish before there was hypertext. There are some other interesting kinds of hypertexts that existed pre-computers. There was this very interesting book called The Syntopicon from Adler that if you want to understand what classical authors had to say about a topic like the father's responsibility to a daughter, you can look that up in the Syntopicon and you will get references across Rousseau through the Bible and so on and so forth. And those are kind of hyperlinks. I mean, they were printed on dead trees.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah. I guess I'm just wondering what the correlation is. Like, if it's the case that actually effective teaching is mostly about empathy, then maybe it's ASCII correlated. Like the people who are going to be good at anti particle physicists are actually, they wouldn't have made good teachers anyway.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Classes, classes that are more about creative ways of thinking with math or open-ended problems as opposed to here's how to do this algorithm because to conduct those kinds of classes, you have to be able to think on your feet. Like you pose a difficult question to which there may not be just one appropriate answer and your students will throw all kinds of stuff at you and you have to be able to take that stuff and integrate it and show how one student's answer relates to another student's answer and show how those conceptions can be built upon in order to produce some useful understanding for what you had in mind. Anyway, this kind of improvisation requires a mathematical familiarity and ability, but I don't think it requires anything like an extraordinary ability.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“I guess what I'm saying is something like if that were true, and there are like a bunch of people who are now, you know, astrophysicists or something, it's not clear to me actually that they would have been the good teachers. Like being a good teacher is often about empathy and effective communication and care. It's very personal. It's very intimate. Like you need to understand the subject, but to teach a 15-year-old or something, you actually don't need to understand it at a like a postgraduate level necessarily. It's very interesting to see that there's a bunch of studies of kind of the impact of domain knowledge on teaching efficacy. I've read some in math. I'm sure they exist in all fields. And one of the things that comes up is if you aren't very familiar or comfortable in math, then you will struggle specifically to do inquiry oriented classes.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Oh, that's interesting. I haven't heard that claim. Yeah, I think I'd have to think about it. I guess it's not clear to me how much intelligence matters, like if you want to think of that as some kind of separable quantity.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Of instruction that are more effective in other ways. It's good to interleave stuff, for instance, rather than doing blocked units where it's like, okay, we're going to talk about the Civil War, and then we're going to talk about women's suffrage. It's better that those are somewhat far apart, but it's better to kind of weave these things into each other, not just in history, but in general. So that kind of dissemination has been happening more systematically in the last few decades. I actually am unaware of any kind of studies or results trying to establish anything about the efficacy of teachers now versus long ago.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“That's right. They're not really responsible for instruction. I mean, I'll say also I know lots of people who have postdoc tutors right now, these people as graduate students, they're very pleased often to have a $60 an hour, say, tutoring kind of commission. And that's a little sad, but the pool of available postdocs to hire as tutors, I think is very large now compared to how it would have been 100 years ago. The pool being bigger doesn't mean that the top 1% are getting more though. So I think that's undecided. There is a question of like, have teachers gotten better at their jobs over the last 50 years, say? And there are some ways in which maybe they have, there have been a bunch of projects of trying to disseminate certain research results, like ways”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Actually, would be the best. There is a bunch of empirical research on tutoring, and one of the questions they ask is, what kind of experience level do the tutors need to have? And it's interesting how far you get in tutoring efficacy when the tutor doesn't necessarily know anything. So just having another warm body there actually contributes a very large effect. No, I mean, things get better as you get an expert. And I also have a kind of healthy skepticism of these studies. Like I think part of the role of having Aristotle as a tutor is communicating a worldview. It's not something that would show up on a test or something that these studies would be measuring. Having an extremely inspiring individual might actually be the important component and inspiring is going to be highly correlated with expensive. I think not necessarily, I don't know. That feels complicated.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“That's interesting. Okay, so that would be an argument that maybe it's not so much about the 20th century that we've gotten worse about this, but more like over history, you know, maybe Aristotle was a tutor to Alexander the Great, and now Aristotle would be like a full professor and wouldn't need to take that job. That might be so. I mean, I think it may be the case that some tutors have been priced out of the market, but it's not clear to me that the most expensive tutors”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Maybe. I mean, so I don't know. Let's see. Here's the theory that seems kind of plausible if someone was going to have aristocratic tutors in the late 19th century, would they now go to a fancy private school? And would that experience now actually be less good for them? I don't know. I think it's probably more likely that they'd go to the fancy private school and also still have fancy tutors and then go to a very exclusive university where they're going to get a bunch of highly hands-on kind of interaction with professors.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“I think it's very hard to raise the ceiling. So the ceiling has aristocratic tutors. The ceiling has whatever family dynamics and heritable propensities produce tremendous intellectual greatness. Early 20th century schools produced von Neumann. And it's certainly not at all clear that they are now producing more of von Neumann's or something like that. In fact, von Neumann's production seemed to have probably very little to do with any kind of mass schooling that we would recognize. As far as the very top, I think that's difficult. We're talking about an institution that was created for, I guess, the masses. I guess that there have always been people who have been using resources outside of those kinds of systems. So the mass system doesn't seem to help those people, I guess. That doesn't seem surprising.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Of the 20th century has been in part one of mass education, where part of why we have a service economy and IT economy is that basically all of our population is educated at a particular level. If you look at there are these national tests of 4th, 8th, and 12th grade math and language proficiency, you'll see really pretty slow movement in the 75th percentile and like practically none at all in recent decades, but you'll see like absolutely enormous movement in the bottom quartile. And so in some sense, the story, especially the last 20, 30 years, has been closing what's often called the performance of the cheap and gap where certain groups, part of underfunded schools or who might have households that are unsupportive or Difficult. We're just not having anything like the educational attainment of their peers. And that story has changed.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Of adults depending on which polls you use are maybe would struggle to perform simple kinds of number manipulation or reading or writing kind of tasks. But our bar is basically moved. It used to be like, can you read it all? And these tasks are like maybe a little artificial. They're maybe not relevant to their day-to-day. And that's actually why they're experiencing this. So the number of people the fraction of the population who graduates at 17 or something knowing a particular amount of stuff has basically moved up monotonically. This is mostly about the bottom portion of the population. It used to be the majority were effectively uneducated in the past 10 or something other than informally and in their trade. And really the story.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Look at the fun answer to your question is actually, there's been a ton of progress. Actually, things are pretty good. One thing is I think the stat is in 1900, 6% of teenagers graduated high school in the US. Now, that doesn't mean that 94% didn't have an education that we would regard as a high school education, but it kind of means that. It roughly means that. Now these people are homeschooled. It's also the case that high school education meant something lesser then. A substantial fraction of high schoolers now study AP courses and complete them in high school. And that's at the high end. On the low end, illiteracy was a very live situation 100 years ago in the US and is emphatically not now. Now, it is the case that something like 10 to 15 years.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Much as is possible. And I think to some extent the student should, or the child should be allowed to feel the consequences of their choices. This is complicated by the fact that, again, I'm not wealthy, but any child of mine would have chances, I guess. If they made some weird choice about a career path when they're 13. So they didn't get into Harvard or whatever. Like that would be okay. They could do. They could be 24 and finally figure it out then, or 32 and finally figure it out then, like it would probably turn out fine. And so this doesn't seem like reliable guidance. You should notice I'm feeling very confused about it.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Is chaining them to whatever that impulse is. It makes them the subject of these tides of impulse. And I think that's a pretty compelling argument. It doesn't authorize tyranny, but it also suggests that, you know, you got to be a little bit skeptical about the planning, the plans of 12-year-olds, I guess. How skeptical should Bone be? I don't know. I think I would probably have stronger opinions on that if I had a 12-year-old. But my instinct as a foolish non-parent would be something like a kind of mix. I would be interested in exposing the 12-year-old to lots of topics and possibilities. I would be voluble in expressing the consequences of any particular actions, like if they just want to compose music all day, we could talk about, well, what does that mean? What kind of life does that look like? I would try to be non-coercive in this.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, this is really challenging. So, one of the heroes of the reform school movement is this philosopher named John Dewey. And he has a lovely book called Experience in Education, sort of written near at the end of his time looking back on all of his efforts to reform schooling in a kind of unschooling-ish direction. He was never as extreme as that. But broadly looking for freedom on the child's part. He makes this wonderful argument that Because these kids, a 12-year-old, doesn't have a fully developed prefrontal cortex certainly doesn't have a fully developed kind of sense of self to let them do whatever it is that their whim commands them to do in any given moment is actually not freedom, but rather is”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“It's actually not clear if you do. My friend Alec Resnick is working on a very interesting school called Powderhouse in Somerville, Massachusetts that does something like the model I just described, where you have adults who are in more of like a coaching role, and they aren't necessarily domain specialists, but they'll connect people with domain specialists. So anyway, I would explore something like that model. I'm sorry, this is a little bit vague. If you want to ask about specific, feel free”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Company which is now defunct, and so this is a fun example to bring up, but it's called Schoolhouse. And the idea was that he noticed that people were getting together in pods. That was the thing we did during the pandemic. And in particular, they got together in pods with their classmates from school, maybe five or six kids. And some of these pods started hiring elementary school teachers who were not working because of the pandemic. And these elementary school teachers would like come to the backyard of one of these people's houses and the five or six kids would get together with the elementary school teacher and like they do stuff all day buying this one teacher's time split five or six ways was actually really very tractable you know say you want to pay the person 50 an hour maybe that seems reasonable for a teacher like this is not that hard to do and actually cost less substantially less than a private school I think schoolhouse costs something like a fifth or whatever the cost of an elementary school once you got to older grades you need maybe specialists”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“And so, yeah, so you're going to hear the foolish response of a person talking about what one would do hypothetically. This is very difficult. School, of course, has many purposes other than instructional, right? It has a social purpose. It has a societal purpose. It has a behavioral purpose. And it also has a pragmatic purpose of basically babysitting. Those things can be unbundled. I think it's pretty interesting to consider that if I actually did have a kid, I would probably consider that project pretty thoroughly. I think it's pretty likely that some kind of homeschooling situation would occur. It probably wouldn't be me being the teacher, but it would probably be the people I would hire. I have some resources. Like, I'm not wealthy, but I have some resources. So that is maybe a difference. But during the pandemic, I was struck by Brian Tobal started a...”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“I think I do want to say the reason why I'm willing to endorse this headmaster's view about the taxonomy has to do with the price. I did a bunch of memorization in high school and it was very inefficient and it was very uncertain. So it was like it was emotionally difficult because I wouldn't even feel confident that I'd learned this stuff. I didn't know what it was to learn something reliably, like to be confident that I'd be able to recall it. And it also, it was hugely time consuming because I didn't have techniques or tools. And now part of why I respond so favorably to like, yeah, just learn the taxonomy is that for me, it was trivial. Like, yeah, sure, whatever. Throw it in the deck, you know, like it'll consume a total of 15 minutes over the next few weeks, and then I'll know it. It just doesn't cost anything. So, yeah, okay, fine. Other things in learning do still have real costs. And those are maybe more difficult to know.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Use willpower to make things happen. So just as an example of something that's totally contrary to everything I was saying, I use a tool called BeMinder, which charges me if I don't do certain things, right? This sounds, I don't know if it's kind of military, but it's certainly more authoritarian than this kind of free-wheeling butterfly's kind of gesture I was making a moment ago. And I use it to make sure that I do my memory practice. Shouldn't my memory practice be so joyful? It's at the center of my research, right? It should be like the most interesting, exciting part of my day, but often it's not. And so I use this to do it anyway. So there's some tension here.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Of course, it is. That's Yeah, yeah, I really enjoy this tension. And I'm probably overstating to the Goggin's reference, but it's a bit of an overstream over correction or something. But this really is how I feel. And I feel this tension all the time. The histories in educational psychology that I'm most aligned with are the most robotic authoritarian kind of histories and also the ones that are most like kind of unschooling and monasteries. I really have a ton of sympathy for elements of both of these directions. And there's kind of a weird synthesis of this in my head that I can't, I guess, fully externalize. I guess part of what I'm saying is aspirational. It certainly is the case that I do in practice”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Like, how many copies of the COVID RNA are likely to actually be in it if the vial is a milliliter large or something? That's like kind of a fun little question. And I can enjoy sitting and noodling on that. And in doing so, I will need to retrieve the volume of the nucleotide to help me make that approximation. So I think there's moves like that that you can use to kind of paper over any remaining stuff that feels kind of necessarily unpleasant or wrote.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“All of these formula. And that still is unpleasant. So, can that be resolved? And I think the answer is yes, actually. So I was thinking about this in the context of the cell biology by the numbers book I was telling you about, where there's all of these things like the volume of the nucleotide I said as a nanoliter. So to study the flashcard, what's the volume of a nucleotide is not terribly pleasant. I'm not sure it constitutes suffering exactly. It's fine. I'll do it while waiting in line. But I think there is a better version of that, which is solving an interesting Fermi problem, which involves that term. So something like if I have a vial of the”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“You're not going fast enough, or like you're failing, or you're struggling, and the misery comes from resisting that. It comes from feeling like you're doing poorly and like you shouldn't be doing poorly. Like it's bad that you're doing poorly and maybe you're feeling fearful, like others are going to judge you or you don't have enough time or something like that. And I think that's basically an emotional problem that needs to get healed rather than like a practical problem with the learning. In the case of something like organic chemistry, where like you truly do just need to learn 200 names or something, one answer is that, okay, that can be done very cheaply using modern memory systems, actually. So like organic chemistry students suffer through this and they don't need to, but even with modern memory systems, like you're probably going to spend a total of, I don't know, call it 100 minutes across some weeks studying.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“I guess I'm asking this honestly, not really rhetorically. Let me try to answer my own question. I mean, let me say first off, I think I am broadly speaking very opposed to what I understand to be David Dogginsque attitudes towards almost anything. In this particular instance, I think what I think is something like, if I ask, why is it miserable to learn a particular subject, the answers that come to mind are things like, first off, I don't care about this subject. And I think that's not what we're talking about. Like you're asking about a world in which these great tools exist and someone's using one of these tools to try to do something they really care about. So another reason why it could be miserable that I think is pretty common is that you have some idea about like”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, I also say part of the trouble here is that the cool shit is very likely to come from students who are performing at the 20th percentile in school because they're disaffected and bored and none of this stuff matters to them. Part of the trouble here is that by opting out of helping these people learn there are all kinds of interesting inventions that could probably occur that aren't occurring. So I don't quite know how to contend with that. I guess basically I'm like trying to bite off a piece of a problem that feels maybe tractable.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, yeah. I mean, that's true. But anyway, I mean, like, I say all this to say, like, I understand why the vast majority of people in education are focused on what they're focused on. And I think it is good, and I'm glad they're doing it, and I'm mostly have decided to let them do that, to focus elsewhere.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“They are motivated by arguments of equity and opportunity. And they want everybody to have the opportunities they had. They're very motivated by the injustices that they see in the differing access and the differing support that different people have. And they're very motivated by the very real disadvantages that accrue to the bottom quartile performing students. It's also true that the marginal impact that you'll have on that student's life will be much greater probably than the marginal impact on, say, an 80th percentile performing student. Or so the argument goes, like that student will be fine, which is like properly true.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“One of our tactics was to lean heavily on social interaction. People like talking about stuff with people if it's like a real conversation and for the same reason that I had to use less willpower to study that quantum mechanics text because you were there with me, a student who's engaged in a real activity with a peer will need less willpower as well. They'll also learn from their peer if you structure things right. So social learning becomes interesting. But I think at a high level I mostly have abandoned this question to others. Basically everyone in the educational space isn't totally true but like you know 90 plus percent of people in the educational space are focused on really like the bottom quartile like not even median and there's a good reason for this like many not quite as many people who are in education”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Take eight, and the rod that represents eight and put two of the rods that represent four up next to it and show that this one you can divide into two rods effectively. But then if you take seven, there aren't, there is no other pair of rods that for the same color you can put it next to it. So you get these different patterns. So things kind of naturally suggest themselves by experimenting with these materials and having conversations with people around these materials. And so one of the things we were interested in was, well, are there things that are like that that are for more advanced topics? Like, can we create something that's kind of like those rods, but that is about a more advanced topic in math or are about debates in history or something like that.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Dynamic media representations of things, the idea being that you've probably seen a movie these plastic blocks or things that people can play with when they're kids to get an idea of numbers and number systems. Kids will play with these things unprompted because they're fun. It's just a pleasure to handle them. It's a pleasure to manipulate them. When you have them in hand, it's very natural to suggest like, ah, can you make a pattern like this? Like why you can't seem to make patterns like that? Why is that? So, you know, for instance, you can start to point out things like quis and air rods is the name for a set of 10 rods that have basically unit length 1 to 10 and they're all different colors. And so you can do things like”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Maybe not from an efficacy perspective, but for me, from like a breaking out of this, getting them to follow goals that aren't their own perspective was to focus on inquiry learning and to focus on transforming the learning experience into something that actually is related to their goals. That is, we're asking questions that are authentically interesting that they authentically want to answer, and that they can participate in in a way that feels”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, they'd be super different. I kind of got out of the educational space in part because I don't like the framing of this problem. Like for the median student, the education system mostly wants to make the student do things they don't want to do. It's not about helping them achieve their goals more easily or more effectively for the most part. It's about achieving goals that aren't theirs. Obviously, that's not always true. But for the medians tonight, I think it kind of is true. I become very interested when I was at Khan Academy, I was kind of thinking about this problem. One of the angles that I found really interesting was so at Khan Academy we were mostly thinking about not just the median learner but like maybe the 25th percentile learner one of the angles that felt most relevant”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“In golf, like you try to get the lowest score. So code golf, you try to solve the problem as minimally as possible, right? Like, ah, ah, you know, I don't need this. I can combine this. I can do it in three lines. If I use Haskell, I can do it in one line. That's a kind of thing programmers do that's like this, but just endless refactoring is another thing that's kind of like this. You have the thing working, but it could be more beautiful.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“It's not the software engineering. It's only designers. Got it. But for the software engineers, I mean, CodeGolf is the equivalent, right?”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“There's this beautiful and very enticing feeling of turning the crank and making it better and polishing it and trying this or that. But when you're trying this or that, all of the options are kind of okay and you're kind of trying them out of curiosity. Or like maybe it can be even better. And that's very different from the kind of design where you're just like, I simply don't know how to do this. And I think it's part of why those designers loved making those posters. It's like it's a snack. It's a treat. It's also something they get to control. Whereas ordinarily they don't.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“So, to some extent, I think they are So the people that I know who are like this, it's a combination of their skillful often and like the nature of the problems that they're solving are highly tractable. So like a kind of an example of a kind of thing that designers will often rabbit hole into is like designing a poster. It actually often used to be kind of a cliche that at Facebook there were all these posters up on the wall of the office very, very elaborate beautifully designed posters for like a talk that someone was coming to give at Facebook. And it's like, why did somebody put all this effort into it? Like, well, it feels really good because a poster is this really constrained, it's like finite, it's ephemeral, you can start it and within a few hours, like, yeah, there's a search space, but you can find a decent part of the search space pretty rapidly. And once you're there,”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source
“Million artboards. They're just trying stuff. And you don't have this feeling, or at least I don't, and I think many designers don't have this feeling of progress. It's like you're just kind of exploring parts of the search space. And you're learning that parts of the search space don't work. And eventually you stumble on one that does, but you don't have this feeling of like getting closer often. Often there will be like weeks that go by without feeling like you're getting closer because what you're doing is just kind of like narrowing the search space.”
2023-07-12 · Dwarkesh Podcast · Andy Matuschak — The reason most learning tools fail · IDENTIFIED FROM THE TRANSCRIPT · source