E378 -

Andy Matuschak joins the show to discuss how different learning models will help students in different ways.


Andy Matuschak joins the show to discuss how different learning models will help students in different ways. They discuss how students best remember material and how we should consider cognitive science when constructing a teaching technique.

What is the purpose of primary school? Why do we group children by age for learning in school? What is the best way to learn from flashcards?

Further Reading:

Experience And Education, written by John Dewey

Quantum Country, by Andy Matuschak and Michael Nielsen


0:00:07.3 Aaron Powell: Welcome to Free Thoughts. I’m Aaron Powell.

0:00:09.4 Trevor Burrus: And I’m Trevor Burrus.

0:00:11.2 Aaron Powell: Our guest today is Andy Matuschak. He’s a software engineer, designer and researcher, he helped build iOS at Apple and led R&D at Khan Academy. His current interest is technologies that expand what people can think and do. Welcome to the show, Andy.

0:00:26.0 Andy Matuschak: Hey, thanks for that introduction.

0:00:28.7 Aaron Powell: Technology has radically changed the way that we seek out and interact with information, but the methods of formal education, the way that we acquire knowledge, haven’t really seemed to change much since the ancient world. Why is that?

0:00:45.4 Andy Matuschak: Yeah, well, it’s interesting the ways that that’s true and that’s not true. When I got started on my ed tech journey, maybe seven years ago or so, that was definitely the story that I told, there’s another common story that the current school system is created in Prussia, and hasn’t changed much since then, and while that’s true to a large extent, it’s interesting also to talk about all the ways in which it’s not true. So for instance, discussion‐​oriented classes being highly commonplace in say primary and secondary education is a relatively new emphasis in the last few decades, and Montessori schools started at the beginning of the 20th century, have been trying to push their way up through secondary school.

0:01:33.3 Andy Matuschak: And so I find these trends promising, I find some of the Common Core’s attempts to emphasize reasoning and argumentation in addition to rote actual procedural knowledge promising. Simultaneously, there are a lot of entrenched forces that make it difficult for these types of reform or progressive‐​minded education policies to move quickly. Everything has to move together, very large institutions consisting of many parts, many different disciplines, from the teachers on the ground to the administrators who have certain expectations, backgrounds and training, to the funding sources for the schools, which are often dependent on certain practices and assessment behaviors to, yes, those assessments, which for the Common Core roll out, for instance, one challenge is that the assessments are not as progressive‐​minded yet as some of the standards in the Common Core, and so teachers are caught between a rock and a hard place there.

0:02:37.0 Andy Matuschak: Likewise, parents can cause difficulty as well as opportunity, either direction, either they can be more progressive in the classroom and trying to drag everybody to the future, but just as commonly, the parents often express concern when their child is learning something in a different fashion than they learned it. They can sometimes feel like their child is getting a short shrift or is not being taught properly, or something like that.

0:03:02.8 Trevor Burrus: Well, we have some weird things that are persistent in at least our public education system and some private schools. One that always has struck me as interesting is grouping children by age, that one, it doesn’t seem to have been re‐​thought much, where I’m like, age does not correlate to ability necessarily. It might, but we just sort of do it and we kind of think that there’s no other way of doing it. I don’t know if you’ve looked into that, is if there’s an alternative way of doing it. We did, of course, used to have one‐​room school houses where 15‐​year‐​olds and 7‐​year‐​olds were together, which actually could have been beneficial, but now we just sort of segment them out, and that’s one of the weirdest things about schools, it’s one of the only places in your entire life that you’ll be with people with your own age as a general rule, that in a sense you don’t really learn about the fact that when you get a job, there’ll be someone who’s 50 and someone who’s 25 and you may be a teen. So that’s always struck me as weird and we’re not really thinking outside of that box.

0:04:00.7 Andy Matuschak: Yes, there are some reform‐​minded schools which are taking alternative approaches there using grouping children by, say, readiness or independence levels, things like this. But one fun thing about this question or concern is that it really starts to access what do you actually view school as being for. And for instance, if you view it as being primarily a psycho‐​sociological developmental institutions, it’s there to help you grow as a person and relate to others effectively, then maybe it would especially make sense to group many ages together in the one‐​room school house cell model, because because maybe socialization is gonna work better in that context. There’s some evidence of that, up to some threshold where things get tricky. Whereas if what you view the goal as, is something like learning a bunch of skills to enable either like teacher vocational work or entrance into the great process of humanism and discovery, then you might wish to group people by, say, knowledge or capacity or something like that.

0:05:20.4 Andy Matuschak: Age is… Actually, an interesting thing that I learned at Khan Academy is that in a typical school that the smear of abilities in a given class is absolutely enormous, and for many teachers, this really is their number one problem. The most common concern or complaint we had been talking to, for instance, middle and high school math teachers, in particular, was the absolutely enormous variation in their students’ capacities, both conceptually, just comfort with numbers, as well as practically, their ability to do certain things, and so they have to somehow teach a class that includes students who are nominally or from some standards’ perspective two years behind where they’re supposed to be, as well as of course there’s others who are a year ahead or something.

0:06:04.8 Aaron Powell: That reminds me of… My wife is an elementary school teacher, and when we lived back in Denver, she taught at a private school for gifted children, and the way that they handle that, because gifted kids vary wildly, was she would essentially write individualized curriculum for each of the 15 or 20 kids in her class, and it was all built around like if this kid was really into hockey, she would build the entire curriculum of math and science and history around hockey. Which is wonderful for the kids, but as the spouse who saw her working in the evenings, it was rather labor‐​intensive, so it didn’t scale as well, but…

0:06:43.0 Andy Matuschak: Yeah, although, I don’t know, maybe it’s not that unreasonable. So there are 55 million US K-12 students. And so if you imagine that a teacher can only support 20 students, which maybe that’s an upper limit for that kind of endeavor, then you’d need 2.7 million teachers to support those students, and maybe that’s not that outlandish.

0:07:05.6 Trevor Burrus: Probably not. One thing I like, I’ve known Aaron for 20 years, going back to Denver. And one thing I liked about what his wife was doing, as a kid who was really into things growing up, and I think probably all of us were, just get really into dinosaurs or baseball, or… I got an obsession with Anne Boleyn, I had some weird obsessions when I was going up, but that kid is sort of… They like to focus a lot on one thing, so she had to… You know, how do you teach math through hockey. Well, that’s pretty easy goals, statistics, but she had to be, how do you teach math if the person is interested in the human body or something like that. But I think that goes to some of your work where if you’re trying to get understanding, but you go through something the person wants to understand as opposed to this sort of transmission model where it’s like so many kids hate school but they love learning about other things outside of school, even if it’s something kind of silly, like Pokemon characters or something…

0:08:06.2 Andy Matuschak: They can recite all 150, but they can’t recite the times table.

0:08:09.9 Trevor Burrus: Exactly, and their stats.

0:08:10.5 Aaron Powell: That was a popular topic in the gifted school.

0:08:13.3 Trevor Burrus: So how do we harness those things together? People have a passion for it. I think even people who don’t seem to have a passion for learning have that. So how do we get those to kind of meet so they can actually get skills, if that’s the point of what we’re doing here.

0:08:26.6 Andy Matuschak: This access is a fundamental rift and challenge in the reform education space, stretching back, especially over the 20th century. The challenge that many people have had in trying to produce, say, student‐​directed learning environments has often been that they don’t quite take it far enough. So one common challenge that we encountered when talking to those types of institutions at Khan Academy was that the teachers really wanted to just support the students in investigating whatever they wanted to investigate, but the district or the parents or the some things would eventually get in the way, and this was true even at the Khan Academy Lab School, which was this very strange experimental school without concrete curricula. One common refrain from those teachers was that the parents will talk a big talk about wanting their kids to follow their interests and their passions. And yet at the end of the day, when they come in for those teacher‐​parent meetings, they’re like, but my kid’s gonna get into Stanford, right? That is a thing that is very important to them. So you have to thread this multi‐​institutional needle, and no-one’s quite figured out how to do that yet.

0:09:53.9 Andy Matuschak: One other interesting challenge is something that John Dewey points out in Experience and Education, which is that, particularly before people have developed their sense of self and the prefrontal cortex, allowing such a person or encouraging them to do whatever they want in a given moment, it’s kind of worth asking, if such a person is truly free, if a person is kind of following their momentary whims at all times without having developed wisdom, is that actually less free than a person who’s kind of gently guided or shown opportunities they might find interesting in a non‐​coercive fashion, but with slightly more constraint.

0:10:40.7 Aaron Powell: A lot of your recent work has been on personal knowledge databases and knowledge synthesis, and I’m thinking about that in the context of what we’ve just been talking about, because that project is almost by definition like cross‐​disciplinary, you’re trying to take a lot of different fields, a lot of different kinds of knowledge, a lot of different directions and figure out how to synthesize new ideas out of them. But the way that we tend to approach education, particularly at elementary level or high school, is to compartmentalize. You’re not learning the way that my wife taught, which is, we’re gonna learn about hockey, and through hockey we’re gonna learn math and science and all that, it’s rather you’re gonna do science and then you’re gonna learn reading, which is kind of an odd subject in and of itself, because reading is just how you approach everything else, it’s not its own thing, and once you’ve learned your letters. That does that kind of trap us, like do we need to break down that compartmentalization of disciplines in order to be more effective?

0:11:47.0 Andy Matuschak: Yes, and actually, this is a place where most of the educational institutional forces are supportive of doing that, so the Common Core actually includes a lot of provisions and encouragement for this to happen. The reading standards, for instance, they include standards for speaking and listening that are supposed to be applied to subjects other than just your English class, for instance. So in order to achieve the English and Language Arts Common Core standards, you actually have to be working on those things in the context of a science class or something, because they include standards around analytical and quantitative thinking. Now, where that’s a trouble is with some of these forces I discussed earlier on the ground, it can be very difficult to do this. As it happens, you have, say, an English teacher and a Science teacher, and the Science teacher is kind of like rewarded and incentivized for the students’ performance on the Science assessments at the end of the year, and the English teacher is not really rewarded or incentivized or even has any feedback around how the student performs on, say, the quantitative speaking and listening, that type assessments.

0:12:58.4 Andy Matuschak: So all the teachers actually I’ve talked to about this have expressed great difficulty in getting anything to change, and this is much, much more true for the Math standards and the Science standards, which are still somewhat more nascent, which attempt to be cross‐​cutting, because those require domain expertise that many of the relevant teachers don’t have either practically or they may have an emotional aversion, so trying to bring number talk into history, for instance, is super interesting. You’re gonna talk about means of production in the 18th versus the 19th century, actually looking at those numbers can really give you a visceral sense for what happened, but many of the people teaching those types of courses have very strong aversions to mathematical thinking.

0:13:50.9 Trevor Burrus: Now, you’ve… This is gonna sound bad, but you’ve come out against books in some ways, or at least you have some issues with some ways that some books are thought to be used would be the more… What’s the problem with especially non‐​fiction books that you have pointed out?

0:14:12.2 Andy Matuschak: Thanks. Yeah, yeah, I should be clear that this is more a provocation than a criticism, I in fact, love books. But I think it’s interesting to look at objects with their purposes in mind, and then to ask how well the object is sculpted to fulfill that purpose. And so if we imagine that an informational text, for instance, that the aspiration is that a reader is going to somewhat reliably be able to internalize a particular perspective or a set of strategies, methods, even just basic facts. It’s interesting how little books actually seem to do as objects, as media, to facilitate that. It’s true also for even things like self‐​help books, if you read a book about trying to, say, form habits better, you might sit down and read it, they’re usually really short, so you read it in an hour and a half, and then you close it and you put it back on your shelf and you go about your day. And then something might happen which would benefit from one of the ideas in that book, and yet the author is back on your shelf at home and is no longer participating.

0:15:28.4 Andy Matuschak: So I’m not really saying this to criticize books as much as to ask, is it possible to invent media forms which can do the jobs of, say, informational texts or your self‐​help book on habit, but which actually, say, participate more deeply or take into account how people think and learn more effectively, things like this.

0:15:48.4 Trevor Burrus: Well, isn’t that a TED talk, it’s like at least part of the point, right?

0:15:53.2 Andy Matuschak: Maybe. I mean, I would argue that a TED talk is also quite ineffective. It’s interestingly different in that it may have, say, like a visceral or a deep emotional reaction, but I’m suspicious how much long‐​term effect a TED talk has on a person. Maybe that’s what you’re saying.

0:16:11.3 Aaron Powell: You’ve… As a way to reinvent the book to, I think, accomplish more of what you’re talking about, you’ve co‐​authored, I believe, a textbook, an online textbook on quantum mechanics. Can you tell us… So kind of use that as a way to give us a more concrete version of what you were just talking about.

0:16:32.9 Andy Matuschak: Sure, yeah, yeah. One thing that’s interesting about very complex topics is that raw working memory can actually become a barrier very quickly. So if you’re trying to learn, say, a complex topic in economics, and chapter 1 introduces a whole bunch of new terms and new approaches, and then chapter 2 starts trying to put those to use very quickly to solve problems, your working memory is really quite limited and you may find that you can’t actually keep all of those things in mind simultaneously, ’cause you’ve just been exposed to them. And this may make it difficult for you to learn complex topics, including those quantum computation.

0:17:19.3 Andy Matuschak: So what do we know about memory? Is it possible to take advantage of what we know about memory in order try to do better? That’s what this book, Quantum Country, is exploring. Cognitive scientists actually understand fairly well how to ensure that you’ll remember something. This is kind of a familiar thing in everyday life, if you hear about something once, then often you’ll forget it fairly rapidly, but you know, if you hear it again, then maybe it’ll last a little longer and then you hear it again, it’ll last a little longer than that. And so ideally, you actually wanna hear about it just as it’s maybe likely to fade away each time, and if you do that, you can actually very efficiently, with, say, only half a dozen reinforcements over the course of a year, keep something in mind quite reliably. It’s just the books don’t really do anything to help you do that.

0:18:12.5 Andy Matuschak: You’d have to be very diligent and assiduous and make yourself a little planner or whatever, so Quantum Country has that mechanism built into it, so as you read there are these little questions, almost reading comprehension questions, they’re very simple, about things that you just read, a little bit like some other online textbooks, but where it differs is that it will follow up a few days later and you’ll look back at all of those things again, and kind of test yourself again on your memory of those things, and for the ones which you still remember, you’ll then test yourself on them again, quite some time later, say several weeks or a month, and then for the things that you’re struggling with, you’ll see them again sooner to reinforce them. And so after a few rounds of this that we found that the readers are able to internalize quite reliably really a very large amount of information from this book.

0:19:03.7 Trevor Burrus: So is this like, is there an app that comes with it, like Duolingo, it prompts you to answer questions?

0:19:10.6 Andy Matuschak: It sends you an email, yeah.

0:19:12.2 Trevor Burrus: Okay, okay. I can see that. I remember back, growing up, there was a lot of discussion, and I’m sure you probably know the source of this, because we have different education theorists who kind of come through at different times, but there was the auditory learner or visual learner categories that came through, which always struck me as a little bit facile. For me, I always thought I was a learner who… As soon as I kind of grocked it, to use the Stranger in a Strange Land, as soon as I kind of got it and I could re‐​combine it in my own head, that was the moment that I wouldn’t forget it. And I’ve always… I thought that when I was like 7, I kinda realized that when I did a, I came up with a totally different way of doing averages that was just different than what I was taught, like it wasn’t complex or anything, I just was doing it differently, and I would never forget it. That seems to me to be what we should pursue, I don’t know how to get there, but anyone, is the moment you understand it, then it’s harder to forget, like really grock it, what’s actually going on. So like is that part of the field that you work in?

0:20:21.5 Andy Matuschak: I would like it to be. This book is designed, and I should say my co‐​author is Michael Nielsen, our approach here is this kind of artificial, informal integration approach, so it’s not relying on you using this knowledge to, say, run some quantum computing experiment, which would be more natural and more pleasant and probably just better if you could make it work. So it’s a very interesting challenge to ask, can you do something which accomplishes the same outcome as Quantum Country for people’s memory without these kind of artificially constructed review sessions. My reference there is games. They are, some of them, the best of them are so, so, so good at this. The real challenge with creating these naturalistic, say, project‐​based learning environments is that you need to structure the projects in such a way as to make use of the pieces of information that the student’s trying to learn, but not too many new pieces of information at once. And so in some fields, the obvious first project might actually require that you internalize dozens of key ideas and you have to find some way to break it down.

0:21:39.4 Andy Matuschak: Because games are kind of inherently artificial, they’re very good at doing this. Games like Portal will have you learned each idea atomically, and the people have broken down that the learning structure in Portal is amazing. So it remains to be seen whether this can be done in general for topics. I’ve actually been assembling notes on this question, but I don’t know how to do it yet.

0:22:02.9 Aaron Powell: It seems like what you’ve described with Quantum Country like that, in order to make that work, you had to write a textbook about a very specific topic and bake in a lot of additional stuff into that textbook beyond the text itself, which is, places a large workload on you as the co‐​author or curriculum designer or whatever else, which then means that for… It’s unrealistic that we would have similar sorts of elaborate things for every potential topic. But on the other side of it, it seems like this kind of thing like, say, using spaced repetition flashcards diligently as a student learning whatever topics it is that you’re exposed to or are interested in, over time, you’re saving yourself a lot of work, like if you can be diligent upfront, you can then, all you have to do is pop into your funky flashcard stack or whatever once a day and just run through them, and over time, the number of cards you have to look at diminishes as you learn them.

0:23:08.7 Aaron Powell: So you have… You save yourself time in the long run, but it requires not just a large time commitment, but almost making a good version of that flashcard deck requires something of a deep understanding of the subject to begin with, so that you’re making the right flashcards. And so it seems like so much of the way that education works now is the teacher gives me an assigned reading, I know that I’m gonna read it, we may discuss it, but reading it is what I have to do, and I can kind of… It’s a concrete thing, and then I know that I’m gonna be tested on it and I can study for that test and then I’m done, but this seems to load a lot more onto the student and in ways that it’s not necessarily possible for the student to off‐​load. Is that a problem for this kind of thinking or is there… Are there tools we can use to help there?

0:23:58.8 Andy Matuschak: Yeah, these are all open parts of the research question. These are good questions. So Quantum Country offers up this kind of trade, and in fact, this is really the core insight of it, you refer to space repetition, that’s the general name for the mechanism which Quantum Country is adapting, and spaced repetition is extraordinarily effective. So it’s worth asking, why don’t people use it? The number of people who use it is very tiny, and yet it really feels quite like magic. And as you point out, one of the key reasons is that it’s very costly, and it is both costly in time and it’s challenging to make these flashcards. It requires both the domain expertise, which you pointed out, but it also requires expertise in this weird medium of making flashcards, which is surprisingly difficult, and it’s not even obvious that it’s difficult. So a very common phenomenon is that people start trying to do this and it doesn’t seem to work very well.

0:24:56.1 Andy Matuschak: And they think it’s because this technique doesn’t work very well and so they give up, but actually it’s because their flashcards aren’t good. So Quantum Country’s core insight is what if the author creates the flashcards and there’s kind of a time trade, so they have to do a bunch of work up front, but maybe you scale it to tens or hundreds of thousands of students, and it makes sense. There’s down sides, making the flash cards actually is very, very good for your learning because you have to do this distillation and synthesis of knowledge and you’ll spend more time in contact with the material and will be more personal and probably connected to things that you care about or you already know about better, but it’s an interesting trade to explore.

0:25:36.1 Andy Matuschak: So then the question is, what’s the right mix? So maybe there’s crowdsourced Wiki‐​style flashcard layers that you put on top of pre‐​existing texts. Or another way I think about it is writing these prompts embedded within Quantum Country, it is still less effort than writing the problem sets which would normally be in a textbook, and yet authors do write their problem sets or exercises in textbooks. So I think the absolute amount of work is not unreasonable, that there’s just kind of like a getting it off the ground effect challenge.

0:26:14.5 Trevor Burrus: Now, Aaron and I both have philosophy degrees, undergrad and law degrees, so with the Khan Academy stuff and the kind of flashcard or repetition, at least where a question can be asked to you, it seems that they’re not easily transported to the humanities in the same way, if you’re talking about large historical periods and understanding what’s going on there. If I got an email that was like, describe the 19th century in five minutes, that would be a little difficult or… I mean, I didn’t make many flashcards in my philosophy classes for my tests, so is that a totally different… Do we have to think about that totally differently and say, this isn’t about spaced repetition, this is about conversations, maybe, so you really get the ideas of Wittgenstein or Plato or whoever, rather than this kind of other learning model.

0:27:06.0 Andy Matuschak: Right. So I don’t know, first and foremost, but I will say that as an amateur student of both philosophy and history, these types of techniques have been quite helpful. They just look a little different than you might be used to. First off, it is just helpful to have simple prompts, like what is sentimentalism, for instance, but that’s not all that interesting. So once you’ve been practicing for a little while in philosophy, you would just know that by way of trying to have debates with people about different philosophers’ approaches to problems. And then you can do these kind of higher order prompts that are more creative or more comparative by, for instance, asking things like compare and contrast positivism and existentialism. There’s kind of some interesting overlaps there, like what are their similarities.

0:28:04.1 Andy Matuschak: And one interesting thing about that question is that you might find yourself actually giving somewhat different answers over time as your understanding develops, and so in a way, this prompt is playing a different role that I’m also finding very interesting, which is that it’s a way of programming your attention. So as kind of a novice philosophy student, the first time you consider a question like the difference between those two worldviews, your answer might not be very deep, and so you’d almost like to program yourself to return to that question, perhaps not in the same very tight memory‐​oriented schedule, but say a couple of times in your first year to see if you have new things to say about it, and so this lets you do that.

0:28:46.9 Andy Matuschak: I’ve also created in my amateur study of philosophy kind of creative‐​oriented prompts with a similar focus on programming myself to practice. And one example would be like, take a recent decision that you made and imagine what you… Decide with your answer to that decision, your choice to that decision would have been if you had taken only a deontological perspective, how would your answer have changed. And that’s very interesting, ’cause it definitionally has different answers every time.

0:29:20.6 Aaron Powell: In this world, though, if we shifted entirely to a world where people were using textbooks like the one that you co‐​authored and were creating decks of cards and programming their attention and all of that, what is the role of teachers? I mean, outside of being the people who write the textbooks and all of that, is there a role for teachers in the traditional sense, in that world?

0:29:45.9 Andy Matuschak: Yeah, absolutely. I’m kind of a partisan in the sense, and so this will reflect a worldview that’s certainly not universally held in the education space, but I view teachers primarily as facilitators, so arranging discussion is I think one of the most valuable things that the teacher can do, and it doesn’t really matter how many flashcards a person’s practiced, a well‐​arranged discussion between them, between experts on essential questions can still generate new knowledge, and so it can definitely generate new knowledge for children or for undergrads. Another, of course, important… Well, for a teacher, particularly for young people, is just facilitating their emotional, even spiritual, psychological development, and then especially in higher ed, an extremely important role that the teacher has is being something like an ambassador for their culture.

0:30:39.8 Andy Matuschak: So if you are a professor of philosophy, one of the things that you are doing is modeling a way of thinking and being in your every word in conversation, and you’re expressing a system of values and norms, and that’s very difficult to absorb. Those are some of the things that I think teachers are still very important for.

0:30:58.6 Aaron Powell: This morning, I was wandering through your notes, which you’ve put online on your website, and I encourage our listeners to do that, because it’s a fascinating experience to walk through these notes and…

0:31:10.9 Andy Matuschak: These are rough working notes, I should clarify.

0:31:15.1 Trevor Burrus: Welcome to Andy’s brain, kind of thing? Yeah, I looked through them.

0:31:15.2 Aaron Powell: I had a lot of fun going through it, but you have… I came across this line that I wanted you to expand on, because both Trevor and I are knowledge workers, we’re scholars, our job is learning about stuff, having conversations about it and writing about it. And so in one of your notes you say athletes and musicians pursue virtuosity and fundamental skills much more rigorously than knowledge workers do, which feels right to me, but I’m not sure why it’s right. Like why that’s the case?

0:31:46.5 Andy Matuschak: Sure. This is happily a question which more talented people that I have also spent some time thinking about. K. Anders Ericsson, one of the founders of the field of expertise, thought this question was very interesting. He didn’t get very far in trying to answer it, I would say, but he gave one simple answer which might explain why, and that’s that it’s very difficult for knowledge workers to do otherwise. There are certain fundamental differences, which he characterizes, between the types of work which makes it very difficult to engage in what he calls a deliberate practice, so the thing that the athletes and the musicians are doing every day is deliberate practice, and a simple way to define that is an activity whose sole purpose is to improve your performance in some particular domain.

0:32:37.0 Andy Matuschak: And so it’s interesting that, we knowledge workers, we do stuff all the time that makes us better, we’re reading papers, we’re writing that book that pushes us outside of our comfort zone or whatever, but the sole purpose we’re doing it is not to improve ourselves, almost ever, whereas an athlete daily is doing activities whose sole purpose is to improve themselves, and so why don’t knowledge workers do well? We don’t quite know how, so some properties of athletics, for instance, and say, violin playing, that Ericsson talks about, one, it’s very easy to very rapidly assess performance, even at a particular sub‐​skill; two, there is specific knowledge about the teaching and improvement of sub‐​skills, so there’s coaching knowledge; three, there is a generally well‐​understood and agreed‐​upon standard of what good is, and a curriculum and so on doesn’t exist really for knowledge workers; and four, there’s a culture and community of coaches who in these fields are previously eminent practitioners.

0:33:53.8 Andy Matuschak: So it would be as if, I don’t know, Robert Caro, instead of writing his next Lyndon B. Johnson book, decided to spend all his time teaching the next generation of great writers or something like that. We don’t quite have that. We sort of do in some small limited cases, and so on. Now, Ericsson kind of stopped there. I think he’s maybe a little more pessimistic than he needs to be. Yes, it’s true, that there aren’t… I can’t specify a complete set of criteria for being a great knowledge worker, but maybe we can agree on things like you need to be able to collect, synthesize, and distill a lot of information effectively and update on that information.

0:34:39.3 Andy Matuschak: Maybe that’s just kind of like a core skill of knowledge work in most knowledge work fields. It doesn’t seem that implausible to me that we could, I don’t know, create some kind of a game where you have to do that or create some kind of a task or test or something where you have to do that, and that you could, say, assess it fairly rapidly or find ways to improve. So I’m a little more optimistic that we may actually be able to identify some of the core tasks and behaviors in knowledge work and possibly then start trying to improve them very intentionally.

0:35:14.6 Trevor Burrus: I want to take a step back or step up to the 30,000 or 100,000 foot view. Now, libertarians are critical of state education as a concept, and I don’t really necessarily wanna get into that, but one of the reasons I find a fascinating question is because it presupposes an agreed‐​upon definition of education, which I think is actually the rub there, where it’s like we don’t… We even, we think we could say the three Rs, but you even find disagreement there, how much arithmetic do you need that we have calculators in our pocket, we need music, we need art, we need global stewardship. And that there’s actually the heterodoxy of humans, which is only increasing, is making it even more of a quixotic endeavor that we can collectively define the meaning of the concept of education. So does that mean that we should be pulling back or at least diversifying, especially as we become more diverse in our education, as opposed to make it monolithic, which is kind of inherent in the state itself?

0:36:24.6 Andy Matuschak: This is a huge challenge and I’m broadly in agreement with what I perceive to be your skepticism of this kind of, of the project of trying to create a unilateral or a monolithic perspective on what education is. Like three primary perspectives, each of which is valid. I discussed two of them earlier, one is essentially personal development, and two is skill development, either vocationally or from a humanistic perspective, and then three is the citizen’s perspective. You need to create a citizenry. And I don’t think I can make a knockdown argument that any of these is right, it really depends on you and your community and your culture and your values, and that’s just the fundamental question of what this whole thing is even for, so even if you were to accept one of those things, like the Platonic perspective, which is one label for this kind of skills‐​oriented perspective, then there’s still tons of reasonable people can disagree style debate about what skills you should be learning and so on.

0:37:29.3 Andy Matuschak: And so the role of things like Common Core mostly plays a factor in the domain of equity, where… I don’t know about that project specifically, but really the project of the education space over the last couple of decades has had just enormous success. So there’s a story that I can tell that has a really sad ending and kind of a really happy ending. There’s this thing called the NAP, the National Assessment of Educational Progress, and it’s been running for a really long time, and every year it collects some sample of performance in mathematics and reading and writing from fourth graders, eighth graders and 12th graders, I think. And it’s pretty incredible to see how little progress has been made, in general, after decades of work in education in, say, the medium, eighth grader mathematics abilities. And it’s especially galling when you look at how much we spend on education, for instance, the combination of public‐​private sector spending is a trillion dollars a year, it’s pretty crazy.

0:38:48.9 Andy Matuschak: But so that’s the sad part of the story. The happy part of the story is that the achievement gap has closed substantially, and you can define that various ways, but basically, if you look at, say, subgroups, whether they’re demographic or economic sub‐​groups that previously lagged behind by multiple tens of percentage points, they’re now lagging behind by, say, single‐​digit percentage points. And most of the story of how that has happened, I shouldn’t say most, a large chunk of how that has happened has been this kind of standardization where it’s like, well, maybe we can’t agree on what education is about, but there are some core things that maybe many of us can agree it should be about, and we can figure out how to do those well and reproduce them.

0:39:31.0 Andy Matuschak: And then the perspective that many of these people have that I don’t necessarily agree with is all this stuff about personal development and finding yourself and pursuing your passions or whatever, that’s all really nice to have when your parents can eat, but maybe if they can’t, then you should focus on able to get a decent job or something.

0:39:53.1 Aaron Powell: I’ve been struck by how much the Internet has kind of changed independent learning in a way that when I was my children’s age, so I was in elementary and middle school, I learned in school and then I would go to the library and get a book, or I would ask my parents something, but that was kind of the mechanisms I had for learning about a new subject. And then there were occasional like educational programs on television, but those were usually pretty lame. And then… But now I watch my kids and they have YouTube, and YouTube has a lot of stupid stuff on it that they open up a Chrome tab while they’re supposed to be in remote learning right now and watch Minecraft videos instead, but they also can… I catch them like that topic that caught their eye, they are just plunging into videos about it, and there’s this extraordinary resource.

0:40:55.4 Aaron Powell: Or there’s people kind of remixing learning, so there’s a woman that my wife watches on Facebook video who does her make‐​up while telling true crime stories. And she just tells these very detailed, 20‐​minute, this is this particular person and how the murders and how they were caught and so on, and she’s doing her make‐​up during it, which is totally bizarre to me, but is utterly engaging. And you can just watch and you’re learning a lot of stuff, and so it feels like a lot of the most interesting stuff in the way that people can learn, especially independently motivated people, is in places that educators or like education community or the way that we think about traditional learning would see as YouTube is this frivolous place where you waste time or this woman doing her make‐​up isn’t like real teaching.

0:41:51.5 Aaron Powell: Is that like a problem as far as fixing things, or is it… ‘Cause I can see it’s a problem on one hand, but it’s also like my kids are spending far more time doing what I would call genuine independent study than I did when I was their age.

0:42:04.0 Andy Matuschak: It’s wonderful. I love it. One thing I think about is, often, is how much I wish I had the 3Blue1Brown math YouTube channel, Grant Sanderson’s channel, when I was a kid learning math, because it is so much more interesting and better explained than anything I had available to me. So there’s multiple factors that people are excited about it, right, so one of them is that, this is kind of a Khan Academy story, maybe you can find someone who is really good at explaining stuff and they might be better at explaining the thing than the random teacher your kid happened to have, who’s drawn from a large distribution of abilities, and maybe the median performance of all teachers is kind of fine, but every teacher is going to have a thing that they are not so good at explaining, etcetera. So there’s this kind of raise the floor type thing.

0:42:56.0 Andy Matuschak: What I find much more interesting is I think what you’re alluding to, and that’s really like a change in focus or purpose. A great thing about the video medium is that it accesses emotional factors so much more easily than text. And so that’s one of the things that I love about 3Blue1Brown, his love and curiosity and passion for the subject just come through in every video, and this is true of MinutePhysics and all these other things. So a challenge that I often hear when I talk to these people, it’s like they’re not quite sure how they should think about what success is, like Grant, for instance, expresses concern or uncertainty, and people watch these videos, millions of them watch these baroque math videos and enjoy them, seem to enjoy them. So what, is that doing anything and do we care?

0:43:52.1 Andy Matuschak: So if it’s the case that all it’s doing is providing entertainment and there’s nothing durable happening, maybe that’s fine, like is it shifts people’s values towards finding that more interesting. Maybe it’s bad and he should try to pursue, say, efficacy to a greater degree, however you might define efficacy. And so there’s a lot of uncertainty in that space, even from the creators themselves. As far as not the creators, then you run into all the problems we were talking about earlier, where your kids are doing this independent study, but how do we assess that on the summative test at the end of the year, I don’t know, so I guess it isn’t valued by the system.

0:44:35.3 Trevor Burrus: In the pandemic time, we’ve seen a lot of things happen with schools, of course, many students not being in school, not getting effectively educated via Zoom, but we’ve also seen people thinking maybe different ways of doing things, different technologies, and there’s always new educational philosophies out there that come and go in different ways, and studies and things like this. I remember when my aunt was teaching in the ‘70s, one of the big fad was this open classroom model where they took down the walls of the classrooms and then they were like, that was a really bad idea, we’re going to put the walls back up.

0:45:12.7 Trevor Burrus: And so it’s always churning, and it’s exciting if you think about it, and this is a libertarian question, of course, is how do we think about the possibilities of what Khan Academy did and all of the different things out there… Aaron mentioned YouTube and so many other things. And then whether or not the public school system can be a handmaiden to that situation or a hindrance, and my perspective is it’s likely to be a hindrance because of all the swirling interests around it. I mean, even just textbook manufacturers in Texas hold an unbelievable amount of power over our public school system and teachers’ unions and all this stuff. So is it something that we should be paring back to encourage this kind of thousand flowers blooming? I don’t know exactly how to educate kids the best, but we need to get out there and experiment, that actually we should be moving in that model, and then some sense public schools are not something we should rely on to be be good for that.

0:46:05.2 Andy Matuschak: I think it really depends on what you value. So if what you value is the opportunity for a higher variance outcome, then I think it’s probably not something many people in the space would disagree with, that, yeah, public schools are not great for innovation, thousand flowers blooming type stuff. I mean, sometimes it happens, but they’re almost like despite institutional structures. And the reason why I think they wouldn’t fight you on that is that they’re basically tuning for a different thing on the landscape, which is they want lower variance outcomes, and they’re maybe willing to accept a lower median as a result. And so if you’re a person coming from a position of relative security or something like you might be interested in a higher variance outcome because you can always do something else or whatever if it fails.

0:46:51.8 Andy Matuschak: I don’t quite know how to think about that, except that if those are two reasonable stances that people can have, then probably both groups of people should have ways to express that, and maybe that’s charter schools for the people who are interested in higher variance outcomes and public schools for the people who are interested in lower variance ones. The challenges here, of course, come in the interactions between those groups and those institutions. There are claims that the existence of charter school programs have some causative effect on declining funding for nearby public school programs. And of those claims, there are certainly instances where that is actually true, instances where it’s exaggerated, so it’s very difficult just to piece this all apart.

0:47:37.4 Aaron Powell: If you look forward 10, 20 years, what does the ideal future of education look like? How do you see… What would education look like a couple of decades from now, if this all worked out and you were put in charge of it?

0:47:58.7 Andy Matuschak: Well, if I had a concise and compelling vision I could articulate, I think that would be a wonderful thing and much, much research would be done. I can describe some properties, I think, of what that world might look like. I would love to see a world in which students are enabled through carefully‐​designed environments that don’t necessarily look like schools, they might look more like what are called Samba schools in Latin America, essentially like community gathering centers, which involve people of all ages. These places contain activities which are naturally appealing to the participants and participating in those activities naturally and inevitably causes enablement.

0:48:52.9 Andy Matuschak: So an example that I find very interesting here in sort of the secular space is Y Combinator. A very interesting thing about Y Combinator is that it has effectively a curriculum, the existence of the structures of, say, office hours and meetings with mentors and the pitch syllabus and so on, all of that amounts to a curriculum, but it’s delivered in this kind of naturalistic fashion, and it’s one where all of the participants are working on something that matters very deeply to them, so I’m very excited about the schools which are attempting to do something like this for K‐​12‐​type education.

0:49:38.5 Andy Matuschak: Ad Astra is an interesting example. The Khan Lab School is another interesting example. So I would love to see more of that. There’s too many unknowns really to be sure. I think the second half and the second main property I’d like to see is a deeper application of what is known about teaching and learning and doing effectively to these environments. So the memory systems that we discussed earlier is one, it’s a fairly galling one, just because the core principles have been known for… Depending on how you count, over a century, or at least many decades, and yet they’re still not part of this environment, and so it just kind of feels like we’re leaving stuff on the table to the extent that kids are struggling to memorize some list of historical facts or vocabulary or the organic molecules that they have to know for the test. That’s essentially just heat loss, it’s just wasted effort. It doesn’t have to be there.

0:50:41.5 Andy Matuschak: But the memory thing is only one example, I think, of a large set of cognitive augmentations, which are possible even just based on what we understand right now. And if we do all those those augmentations, then I think they will help us understand much more.

0:50:55.1 Aaron Powell: Thank you for listening. If you enjoy Free Thoughts, make sure to rate and review us in Apple Podcasts or in your favorite podcast app. Free Thoughts is produced by Landry Ayres. If you’d like to learn more about libertarianism, visit us on the web at www​.lib​er​tar​i​an​ism​.org.