I will add new choices every friday. I didnt want to add all options beforehand because the roster isnt fixed yet (for example, we got this honorable review turned into a contender this week).
If you want to bet that the winner will be neither of the current choices you can bet on "Other" and your bet will be converted to any new choice added in the future automatically
Gotcha. My intuition says that the odds for the first two reviews will be a lot higher now, just by lack of specific other choices at the moment. I've read all the finalists already, and I still think AlphaSchool will win, but I think I can get better odds (in the sense of lower probability and therefore more margin) later on.
Are all the finalists listed somewhere? I've probably read all the finalists, because I read most of the review submissions (and almost all the reviews which were not of works of fiction,) but I don't know which made it to the finals apart from the two which have been posted so far.
For what it's worth, I had one specific frontrunner among my votes, and it wasn't either of the ones posted so far.
It looks like all my top-scoring reviews made it onto the list, but there are a few there which surprise me, which made it into the finals or honorable mentions despite my giving them unexceptional scores.
I don’t think schools are that good at creating strong motivation but it’s true that they provide an overall structure that semi-forces many kids to learn something. However I think we’re at the point that we can offer the equivalent of one-on-one tutoring for every student. “Unstructured learners” can learn directly from books, but better personalized learning along with a supporting social structure can help everyone.
The issue will be how the personalised learning forces kids to complete as much as they can (instead of confusing lack of interest with being stupid and not making them do much). "Motivation" is being used in slightly the wrong way in this essay; many (most?) children don't want to learn whatever the school's trying to teach them, and if given the opportunity to slack off they will (these are the "low structure" kids); school motivates them by forcing them to sit in a classroom and not focus on anything else, but it doesn't make them more motivated in general. With personalised learning, they'll learn at a snail's pace because they don't care about advancing through the material. Or the personalised system will declare that they should be working faster based on [brain scans/genome testing/extrapolation from some kind of reasoning test] and force them to learn at a somewhat arbitrarily faster pace, in which case they may as well be in a classroom.
I hope that personalized learning will not only mean different speed (although that alone is already a huge improvement over status quo), but also different choice of topics.
What I would like to see is some core "everyone needs to know this", followed by tons of optional topics. So that for an average child, 50% of school would be optional topics, and for a gifted child maybe even 90%. (Every year, you would have to gain at least X core credits, and at least Y total credits.) You should totally have specialized topics, such as "classification of dinosaurs".
The advantage of computer teaching is that those topics could *accumulate* over time: maybe the first year there wouldn't be much of a choice, but 10 years later you could have tons of optional topics. (The system should be open, something like SCORM.)
Motivation is the parent's job, not the school's. If a kid doesn't want to learn then he's probably gonna have a bad life. You'll do more net harm wasting bureaucratic resources trying to reach that kid than you will by shoving him out of the way so the smart kids can learn.
You're talking about the vast majority here, would you prefer them to get no education at all, the whole school system dedicated to the 5%? Can't we do it the other way around: a school system for the majority and one-on-one tutoring or online learning for the motivated 5%?
The vast majority already don't get any education at all. I say we stop pretending that they are. Total government spending on education is in the neighborhood of a trillion dollars per year, and most of that is on maintaining the kabuki theater that everyone can learn. Just accept that it's not reasonable to have our current education goals. Optimize for the top 20% because that's who makes the economy go. The rest can be tracked into low-cost high-efficiency warehouses where modest but achievable goals can be hit: learn to read, basic arithmetic, some cultural indoctrination, and maybe teach a trade. Everything else is just a waste of taxpayer dollars in my view.
You are almost entirely wrong that the top 20% is what makes the economy go. Our economy is built on the backs of relatively cheap educated workers. Schools do such a good job of this that these people are invisible to you. But read back to the post-war period when all this stuff was booming -- it wasn't being taken for granted then.
That may have been true in 1950, but in my view part of the reason for the GDP/median wage decoupling in the 70's is that it's increasingly not true. We have an increasingly complex economy and fewer and fewer people can meaningfully contribute to it. I actually think that 20% figure is generous: it's the entrepreneurs and innovators in the top 5% (or 1%) who really matter and that is reflected in the increased share of total wealth that they now hold relative to 1970. The "backs of relatively cheap educated workers" are actually the other 15% of the top 20. Below the top 20% or so and you're not actually dealing with knowledge workers. They may have a mediocre college degree but they're essentially just moving boxes around, answering telephones, filing paperwork, etc. They actually don't need the education, it's simply an aspirational status marker. That's why real salaries for non-STEM degree holders haven't really moved in 40 years.
I think the 20% number is being generous here. Economic growth comes from innovation and technology driven by the top of the top and (to Michael's point) implemented by the majority who need to be able to read, write, do sums and carry out the countless manual and interactive tasks with a smile.
Agreed. My point is that the white-collar hoi polloi below the top 20% don't need any real educational training apart from basic literacy and numeracy, and that can be attained much more efficiently than it is now.
I was tripped up by that too in an otherwise very polished review. After learning the term, I decided maybe in some contexts I'm a low-structure learner, such as this book review contest. I hadn't read this review earlier; in fact I hadn't even noticed it; in fact I read only a handful of the reviews; in fact I could scarcely get through glancing at the titles of the reviews because the structure of the Google doc was a bit too low for me. Now I've read it precisely because it was presented for all to read, today's assignment. I don't have much to say on the subject, but will get enjoyment out of the responses others in "the class" share.
Interestingly, school itself did not function this way for me.
I think this was great read after the previous review. I don't know if the order was intentional or not, but the contrast makes it a perfect companion.
> You might think that we’ve found the solution to tracking. We just need to get all the no- and low-structure learners together and let them move much faster. Here’s the issue. The no-structure learners will always be bored, as long as we are committed to putting them into classrooms where everyone learns the same thing. And those classrooms where everyone learns the same thing are exactly what the low-structure learners need. As soon as you create a higher track there will be a ton of demand for it. Parents will insist that their kid join. And as it grows, it won’t be able to accelerate very quickly. You still need the structure of a classroom where everyone is learning the same thing, and that just isn’t a very efficient way to teach.
I think this might be the fate of every innovative school that tries to serve the unstructured learners. As a school suited for the top 5% of unstructured learners starts attracting a more general student body, largely due to its success at producing high-performing students, it will be forced to adopt more generalized structured learning to accommodate for those new students. Administrators will justify every step as simply adjusting their teaching style and structure to work for larger groups, and at the end of the day you end up with a different, but not revolutionarily different structure to what was already available in the school market.
Maybe the only solution is an eternal pattern of innovation --> decay --> innovation. At least it gives us something interesting and difficult to worry about! Or maybe AI will genuinely be transformative, allowing for extremely tailored learning that identifies what style works for each student, and gives them more of that. Such a thing would only be possible by throwing large amounts of resources at the problem, which is impracticable now.
Anecdotally, Montessori motivated my children much better than public schools have. In Montessori, in particular, children don't all learn the same things. They are encouraged to work on the things that they want to. All children are taught the same works, but what children work on are largely self-motivated. This worked quite well for us, and also worked very well for the parents that I speak to for their children.
It would be valuable if you could describe in more detail how the Montessori system works (in practice, that is). My first question is what is the student/teacher ratio? It seems intuitively plausible that if a student gets plenty of attention from a teacher constantly assessing what the student is motivated about right now, and carefully feeds them all of the curriculum over time, that would work well. But it wouldn't scale.
There’s a lot of information online and I’m not an expert. (Also any school can call itself Montessori so likely a wide range in what schools actually do). Relatively large classes, with (usually) one teacher and often one assistant, student ages span three grades. The room is full of different ‘work’, and the children are allowed to choose what they do but only after they have been trained how to use/do a particular piece of work. Each piece of work often has different levels, which the kids progress through. (Search Montessori beads for an example of math work). Older children often teach younger children how to use certain pieces of work. I think lots of folks think that Montessori students have more freedom than they do — it’s a highly structured environment that nonetheless allows the children to make decisions for themselves within that structure. Lots of emphasis on using and putting away the work in a prescribed way. Like most private schools it ends up being very selective in who attends and who teaches. I also don’t know anything about how it operates past third grade (US). Most Montessori schools stop at third grade or earlier. My sense is that it works well for no, low, and high structure students described in this review because it is in a classroom with lots of structure.
There is such thing as certified Montessori school, but most "Montessori schools" out there are not.
> Older children often teach younger children how to use certain pieces of work.
This is one of the crucial components missing from traditional schools. Kids can teach each other... that is, if you allow them. Some older kids will be happy to gain status by teaching the younger ones, and many younger ones will be happy to get attention from the older ones.
Traditional school bans this, and then people complain about the ratio of teachers to students. Well, it's the school system that made the ratio artificially low in the first place! Sometimes the easiest way to get out of the hole is to stop digging.
I think the obvious question is, if you delegate teaching to people who are only limitedly accountable to you, how do you ensure it's happening properly? I vaguely recall that the biggest obstacle is that supervision costs make it fail to save teacher attention on net.
One possibility is to alternate classes taught by a teacher with classes taught by older students (kinda like at university: professor and assistants). The teacher will notice if the kids keep getting something wrong.
You could also alternate the older students, so that if one explains something incorrectly, the other has an opportunity to fix it.
A different possible approach is to have computer exams that ask you whether you have received instruction from another student. If you pass the exam correctly, the other student also gets a small reward. But if many students taught by the same student keep making the same errors, the question is added to older student's spaced repetition schedule.
Strongly agree with this comment as I was thinking of my children’s Montessori experience while reading this review. I would add, though, that Montessori is highly structured it’s just that much of the structure is hidden from the children and done in a way that allows the children to work independently and to choose (within in a carefully designed structure) what to work on and when.
Yeah, maybe this was one of the iterative improvements possible within the well-structured school environment. We just need to teach public schools how to structure the learning in slightly different ways. Encouraging high agency is definitely a goal, not just motivation while in the classroom.
I'm skeptical if only bc Montessori is significantly more popular for 3-6 year olds than teenagers. If teenagers were more suited for it, I suspect this would have shown up already in the existing market for schools.
It just seemed to me that apart from shop or atuomotive class or home ec, the interior of a school - [and my fairly brutalist in effect if not in design middle and high schools dated from the advent of central air-conditioning in the South so had virtually no windows, and none in many classrooms, which I admit confounds the issue for me] - is not a great way to interact with or learn about the real physical world, nor about one's interests and possible aptitudes within it.
It seems to me that a high school chemistry or physics lab -- even a windowless one -- is a vastly better way to learn about the real physical world than what you seem to be envisioning.
Montessori schools have a high selection effect. They tend to end up with a lot of the no-structure and low-structure learners, because those are the students who succeed in that sort of environment. High-structure learners are less likely to be enrolled in a Montessori school, and less likely to stay once enrolled.
Very possible. Could the "gifted" program at school be cross-grade, and more like Montessori, then? The gifted programs I know of still segregate by grade, and the one for my oldest son is even more isolating (they are given study materials without even changing their classroom)
My child is self motivated. At public school, he is not motivated. At Montessori he was. This is a sample size of 1 so don't take it super seriously, but..
When I talk to other parents, they have similar great differentiated experiences with their children in Montessori. It's a great structured environment that teaches agency and independence. Highly recommend if you can do it.
I suspect public schools can learn at least something from Montessori schools.
I appreciate this review, but there were a whole lot of places where it was just stated that some particular intervention can't be done or won't work without any evidence. I noticed this most strongly about tracking. Tracking apparently doesn't work or can't be done...until high school I guess where it is much more common (and apparently works/is a good thing?)
This feels to me like someone working backwards from a conclusion and not actually being interested in solving the problems that even they admit exist in todays schools.
It's an interesting reflection to the alpha school review which was incredibly positive, to the point of probably overselling. I see these two reivews as the two sides of the school debate coin.
One side is convinced that large improvements are everywhere and ripe for the taking, and the other side things that improvements are the next thing to impossible.
Many of those points do have links attached. E.g., in regard to tracking, the text is "In the US this is typically called tracking or ability grouping. It’s a complex and controversial topic. The research base is hard to read because there are a lot of ideologically motivated researchers who are either for or against tracking and want to see the evidence a certain way. But the biggest theme in the research is that the effects are small. There are plenty of meta-analyses that find an effect near zero (here’s one example)." and the link is https://journals.sagepub.com/doi/10.3102/00346543221100850
The problem is that the review isn't a literature review, it's pitching a thesis on why school has to be set up the way it is, and is trying to persuade readers of this by telling an epistemically-satisfying story that they can internalize. It does some parts of this successfully, but when it comes time to explain why a tracking-based system doesn't work, it instead retreats to throwing out a citation to a meta-analysis that says tracking in practice doesn't currently shift test scores much. But it doesn't really explain *why* that's the case (our social science isn't yet good enough to answer that with much confidence), and certainly doesn't make convincingly clear why its thesis implies that tracked classrooms can't work, because it's not obvious why they can't provide peer motivation.
New idea for a startup: instead of training AI to be great at teaching kids, train the AI to be great at motivating kids to learn. Make it create a psychological profile on every student, and do whatever it takes to convince the kid to learn. Some kids will be paid a dollar for every book they read, some kids will be put through stimulating educational games, some kids will be shamed into conformity, some kids will be put against each other to compete for high scores, some kids will be spanked by a robot.
I think the essay strongly implied that AI could never be great at motivating low-structure and high-structure kids. They need the presence of peers doing the same thing.
In other words, "personalized learning" just won't work for most young people.
The essay acknowledged that 1:1 tutoring works, it just didn't dwell on that because you obviously can't do it for every student if the tutors have to be trained humans. Human-level AI could change that.
Maybe I'm assuming too much but it seemed to me that he thought "human-level AI" could never work as well as a "trained human" because, well, it just isn't human. There's something about an actual person that is necessary. I suppose I'm reading between the lines of passages like this:
"One form of learning that has been shown to be particularly effective is deliberate practice. Deliberate practice involves being pushed outside of your comfort zone, focusing on specific, concrete goals to improve performance, and getting consistent feedback. One common characteristic of deliberate practice is that it isn’t particularly fun. Most contexts where deliberate practice is common, like sports and music training, involve expert, individualized coaching. The coach is mostly there for motivation. The coach does other things as well, but the most important thing a coach can do is motivate you to train. Deliberate practice isn’t common in school learning, but it’s a good reminder that motivation is the key to lots of forms of learning in and out of school. Learning isn’t always going to be a ton of fun. In the absence of one-on-one tutoring for every student, conformity is the best tool we have to create the motivation necessary for learning."
I doubt this is because motivation requires learning from someone with an immaterial soul. It seems more likely to be something like, a human tutor actually pays attention to whether you're learning or just tuning them out, and can respond accordingly. Digital learning systems that have been deployed at scale thus far can't do this.
That's pretty much what I was thinking. A human can and eventually will tell you "Don't waste my time", while an AI can't do that convincingly.
I suspect this is also why coaching works to the extent it does. Both that the coach will say "Don't waste my time" and that you will get cut from the team if you do.
I'd also think that a human tutor can better serve as a model of behavior in a way an AI can't. A human child can aspire to be as good as or surpass their human coach, not as easy for their AI one. Operating words here "not as easy," not saying it's night and day or impossible.
We will have to make some radically retrained LLM's if we want to be able to shout at the kids. But I guess that anyway is something we need to teach them before they can become middle managers. Might need to unlearn the spanking bit from the manager bots, though...
Really interesting article. The motivation issue is really what things boil down to when you look at any content online about learning, beit coding, math, music, art, language... the most popular videos on YouTube are all about issues with *sticking with it*, with not *picking up your phone again* instead of doing the things you want to learn.
I don't think learning most things is very difficult if you genuinely put two hours a day of some combination of actual concentrated practice (doing the thing) and study (learning why you still suck at the thing and what you're doing wrong).
There are some smaller issues with plateauing / dunning kruger if you do nothing but practice and zero study-- you end up in the Tim Buckley zone, famous webcomic artist who copy pasted the same lazy face onto all of his characters for like a decade. I think webcomic artists so often plateau in ways that are easily lampooned because they are stuck on a regular schedule to keep their fans happy, and they already have a dripfeed of people telling them with every update that they're already wonderful and don't need to improve.
If you do nothing but study, you have the 'swimming on a tatami mat' issue, you can watch a thousand hours of painting theory videos and not get any better at painting.
(Also, language learning is different and doesn't work how most people think, you need to spend like 80% staring at target-language subtitled telenovelas or anime you can barely understand, or scraping through easy reader books aimed at toddlers, and you have to give these boring activities as close to your full attention as possible for hundreds of hours, so that your brain builds an intuitive understanding of how the language sounds and works in normal situations.)
But as long as you pay mind to those potential pitfalls, **the big big big issue is motivation.** It's *getting* that 2 hours a day of study and practice.
And the only silver bullet I'm aware of, when it comes to motivation, is having a desperate *need* to learn something, a 'sink or swim' type situation where you literally have no other option but to do it (I think this is the environment that Tiger Moms try to cultivate for their kids, and it obviously often backfires when they get older and realize that everything is made up and the points don't matter).
Language learners do this without realizing it as kids when they grow up in a new country, or want to interact with the majority English speaking Internet and just learn by necessity as they spend hundreds and then thousands of hours in that environment.
If you don't have that desperate *need*, then the only other solution is to make sure you have a significant amount of tolerance for frustration and feeling 'low status' because you can't draw a circle or watch an episode of Crayon Shin Chan (think Japanese caillou but funnier and dirtier - yes, it is a kids show) without English subtitles, and make a real serious habit, OR, as the review describes, find yourself a group of like minded peers around your skill level and stick to them like glue, keeping each other accountable.
Otherwise... learning is fucking hard. We've got phones and porn and video games, why would you spend 2 hours a day fucking around in Blender or drawing badly or watching anime you can't understand, making tiny incremental progress towards some imagined goal 10,000 hours in the future that won't feel anything like you imagine it will, once you actually get there?
Interesting that despite coming from almost exactly the opposite perspective as the Alpha school review, they actually have the same conclusion—that secretly the key to school is motivating students. I liked the distinction between high-, low-, and no-structure learners. I’m definitely a low-structure learner, who has always kind of beaten myself up about not being a no-structure learner. Put me in any kind of class and I will do all the assignments and quickly learn the material. But when I try to learn things on my own from pure interest I always peter out after a few weeks at most. The times I’ve had success have usually been when I signed up for an actual class with teachers and other students. After reading this review, I feel like I need to just accept this about myself and sign up for more classes when I want to learn things.
I'm guessing that you mean that students could cheat with ChatGPT, and that would remove the incentive to learn the material? If I'm understanding the Alpha school approach properly, I don't think that cheating that way would be easy. ( _Is_ that what you mean? ) If I'm understanding the Alpha approach correctly, (setting aside the home school option), there were very frequent quizzes in the classroom, where cheating would require having access to to the LLM in the middle of a quiz. If the kids are required to put away their phones during the school day, and not fire up an LLM tab in the middle of a lesson+test, this should be feasible to prevent.
> they actually have the same conclusion—that secretly the key to school is motivating students
The difference is that the Alpha school review concludes that kids should be motivated by learning opportunities and rewards, while this review concludes that we need to sacrifice the learning opportunities of the no-structure learners to motivate the high-structure learners by their presence. The effect on high-structure students may be comparable, but the effects on no-structure students are dramatically different.
Yup! And, bluntly, the no-structure students are _important_ (making the assumption that AI doesn't totally take over). They aren't _all_ of the future decision-makers, nor are _all_ future decision-makers no-structure students, but I expect there to be a lot of overlap, and, for our society as a whole, it help a lot if future decision-makers are well-informed.
Great review! This crystallized my own observations about school and my place in the rationalist community. The rationalist community is filled with no-structure learners like Zvi Mowshowitz and Eliezer Yudkowsky, and consequently much of the prevailing wisdom doesn't work as well for low-structure learners like me. For instance, I spent a while trying to read textbooks on my own like LessWrong perennially suggests, and I made some gains but lost interest once the material got uncomfortably difficult.
One thing this review doesn't go into that I definitely observed in my own education is that accelerated learning can be pretty traumatic for someone who's used to not needing much structure and is suddenly thrust into a situation where they need to avail themselves of all the structure just to keep up.
The review is great at describing the needs of the learners who needs structure. But the suggestion that the no-structure learners should be sacrificed to them is triggering as fuck.
I need to calm down, and maybe in a week I will be able to write a reply that will consist of something other than screaming...
I believe Scott has written at least two posts about school that I've read. Between those and the discussion it spawned it my impression was always that many around here (and on reddit) experienced the downsides of that rigid structure and none of the upsides. They then concluded that obviously compulsory schooling must be bad and evil, no qualifiers.
My own experience was markedly different. I have, at many points in my life, required external motivation for something I knew was good for me but that I couldn't motivate myself do. Much more so when I was younger still. I'm extremely convinced school would've been in the category "I should do that, it would be nice, but, oh videogames!". Looking back I'm very happy I was made to go.
A lot of the discussion around here has been around largely self-motivated people unhappy with how much school got in their way. I'm happy this review directly addresses that this is true, it does exist, but it's not malice, it's a tradeoff. And then goes on to explain that in detail.
Already waiting so read all the rebuttals and why, actually, school is just a socially acceptable way to imprison small humans who don't have the power to stand up for themselves.
Education is one of those things that everyone has an opinion on, because we were all educated in some way, and for the vast majority of people posting here, it was probably in schools. And as you point out, the typical posters here are not at all representative of the median American.
The simple fact is that compulsory education isn't going anywhere anytime soon, because parents need childcare. Unless an AI robot tutor/nanny can stay home with a five-year-old, it makes sense to have schools for kids to go to while their parents work.
Nannies aren't required to know anything needing teaching, just to supervise children. This will always make them more efficient when the goal is just to supervise rather than teach.
I think the "meeting student where they are is a failure" part needs its own article/review, because it's repeatedly relied on by this article but I don't think it has a strong foundation at all. Also it doesn't seem to pass my priors. From my own experiences, it's obvious that grouping students that have the same characteristics would be better than having too wide differences. It makes me starts to skim this review to get to the point fast, unlike previous review where 18k feels short. I guess the conclusion is generic enough that it shouldnt be 100% wrong, but I don't think the arguments really support it or stamp out denying questions.
I liked that the author put the thesis near the beginning. I'm not quite sure why—I think it helped me read better somehow. I wish more people would do that.
It’s a basic principle of good writing - put the thesis near the beginning, so that you can read everything in the context of where it is supposed to lead.
Scott gets a lot of mileage out of not following this principle. In many of his best articles the reader is being taken on a journey with no inkling of the destination.
One tangential point is "The common school movement emerged between 1830 and 1860." That's about the time of the Jacksonian Revolution, when the political philosophy of the United States went from a "republic" where the voters were the highest social classes and expected to wisely choose for everybody to a "democracy" when the voters were (conceptually) all adults. (Of course things weren't so pretty in practice, but there was a substantial broadening of voting, moving away from the idea that you had to *qualify* in some way.) There was also a high level of immigration. So getting everybody to think in generally the same ways about *everybody* as the body politic was starting to become important.
I like this review a lot. It does a good job arguing against the Alpha reviewer’s (even at the time obviously) crazy claim that the new system could work for 30-70% of students. But it does very little to persuade me against the weak-Alpha argument, which I read as mostly “hey, look at this awesome new system that takes the top 5% and helps them learn twice as fast.”
Honestly, I don’t think the author of this review would have any problem with that claim! For no-structure and maybe even some low-structure students, it’s a really excellent model. But this is definitely a good reminder of the commonsense truth that many people just suck at learning, and old-fashioned schools are set up to teach even them to read.
All that said, I think this author is too gloomy (or maybe just not excited enough?) about increased tracking. It’s true that many high schools already give talented students some opportunity to jump ahead, but it really tends not to be much, and there’s certainly a prevailing attitude where parents and administrators try to push anyone into advanced classes they can.
I’m not sure my experience generalizes super well, but I can tell you there’s at least one school where:
- Everyone takes the International Baccalaureate curriculum.
- Entry to the school is decided by lottery.
- Some students are encouraged to drop out when it becomes clear they can’t handle it.
- Many (*many*) others are not.
- Classes move at a snail pace and no-/low-structure learners (*cough* me and most of my social group *cough* maybe 10-20% of the school *cough*) get bored and play Geoguessr in Physics.
- The teacher doesn’t care because the smarties are still scoring at the top of the class on all the very-slow-moving quizzes.
- IB exams roll around and the school performs *terribly.* I mean, seriously, one of the worst in the world. We have *never* had a *single* student receive a 7 in Global Politics.
- Even the best students tend to be unprepared. We haven’t got results back yet, but I’m pretty sure I shat the bed on Chemistry and maybe Physics, and I have (genuinely smart and talented, way beyond my own ability, etc.) friends who shat on Math, Physics, Spanish, Biology, …
This is terrible! Obviously there should be more tracking and separating done, and it’s true that at many high schools there is, but there also are *plenty* where this is the story. AP/IB classes full of confused, slow, distracted, and distracting learners who mess things up for everyone else.
And even if most high schools aren’t like that, nearly every middle school is. I’ll admit it matters less in elementary, where everyone’s fine just playing with blocks or whatever, but most smart young people I talk to have middle-school horror stories. Bullying sometimes, but mostly just rowdiness, unpleasantness, loneliness, and boredom. School as an institution is probably good on measure—but that doesn’t mean we can’t make it massively better for the kinds of kids who’d be into Alpha.
Even just based on my own experiences as a student I think I agree with a large fraction of the claims made here. But, there doesn't seem to be any attempt to account for the costs of the incumbent approach, or at least there seems to be an underlying assumption that the costs are worth paying.
One is the cost of holding back the students who could be learning more. What would actually happen, in aggregate, if we let some fraction of students fail to learn anything, in exchange for letting some other fraction of students learn several times faster? What would happen if we let students choose to ignore (and fail to learn) subject A, but excel far beyond 'grade level' at subject B, until such time as they find a reason to put in the effort in subject A? I have no doubt this would cause some problems, but it would also do things like create a highly motivated population of scientists and engineers and doctors and lawyers and artisans whose careers are 10-20% longer than they would be under the current model, and they'd enter their most productive years with an extra 5-10 years experience under their belts.
Another is the cost of causing mental and physical health problems. It's a cost to force kids to sit still for hours in a row, and restrict their ability to go use the bathroom or move their bodies, and to crush their desire and motivation to do various things and explore various interests that would, for them, be valuable. It's a cost to force kids to be in the same room as people they can't stand or who treat them badly, for months or years. It's not clear how much, now or in the past, the loss of agency school demands is worth the extra learning some proportion of students will get in return.
And of course, there is the direct financial cost. K-12 schools employ 8 million people in the US. How many fewer teachers and staff would we need if most schooling were replaced with other forms of instruction, AI or MOOC-like otherwise? At the cost of how much learning, for how many students?
Relatedly, what would be the learning *benefit* of such replacement be, if it enabled people of any age to choose to start learning any subject at any time, without fear of social sanction for seeming to be 'behind' or in 'remedial' classes, or needing to go a bit slower, or wanting to study something unusual? What is the cost of us having this societal idea that a student is something you *stop being* after a certain age, unless you go 'back to school,' at a fairly high personal cost - of seeing education as a separate magisterium from the rest of life?
I've spent the last few years RVing full time. I don't have kids, but I've met (and read/watched content from) a lot of families that roadschool their kids. What you can't avoid noticing is that 1) They tend to sign each kid up for different forms of learning depending on their personality, because *that is what best motivates each kid to learn.* There's a lot of options out there these days. and 2) The kids themselves seem to be more mature, independent, and worldly on average. But also 3) Sometimes the kids express a desire to go (back) to regular school, and when they do the parents should, and often do, listen.
I agree strongly with this comment, and I disagree strongly with the fairly common view that because high performing students will do well in school and life anyway, it's okay to waste 70% of their time in school and several years of their life on a system that isn't designed for them.
To a small extent, the UK is closer to a personalized system (and most of the commonwealth countries have the same system). Students are placed into specialized tracks of their choosing at age 13. And people complete their law degrees, medical degrees and PhDs on average 2-3 years earlier.
It has it's costs though. It's very difficult to switch tracks and students are less well rounded. I knew some brilliant science students who didn't know what apartheid or communism were.
The U.S. public schools spend roughly 3x more per pupil today than they did in the 1950's, with no clear evidence of better results. So even if we don't tear down the fence, maybe we don't need to maintain such an expensive fence.
I know people who, thanks to tracking and gifted education, were able to graduate high school a year early or finish college in three years thanks to AP credits, or both. Isn't this a good thing.
I was disappointed that the effect of parents was never discussed. I still remember working with my mom regularly through high school; she provided detailed comments and corrections on every paper I wrote. I did the same for my daughter and included math / sciences in the interactions. We both became very successful technologists. Friends tell me that the parents who attend parent teacher conferences typically have successful kids. Perhaps we need to improve the education results by making the parents more responsible for success?
Kids don’t get to play football unless the parents do certain things? Require volunteer parent activities? Recognition for outstanding involved parents?
Some of it is just expectations. I think the emphasis is on “schools must educate the kids” rather than “parents must educate the kids and the schools will help”.
The parents you need to motivate are the ones who aren't already very invested in their kids' academic success. None of those ideas sound like they would work on those parents.
If you think it’s hard to motivate students, who actually have to be in the place where you’re working, it’s going to be even harder to motivate the parents, who spend almost no time there.
One additional thought on why MOOCs have such a low completion rate: I started listening to one about ten years ago, and the first ten minutes was the professor covering minutia about attendance, grading, etc. It was literally just a filmed lecture with no edits. I very quickly returned to buying college lectures from The Teaching Company.
What if the top 5% were taught to maximize learning and the rest to maximize motivation? The layman can benefit a lot from civility and motivation. But one Einstein can change civilization.
The education system in America is socialist and oddly enough public schools at communist countries seem better at identifying and supporting outstanding academic talent at an early age.
Such a system for public schools was tried in Britain in the 1950's. Students scoring high on tests was packed into special schools around age 11-12. I think there was a huge back clash when upper class parents realized that not all their children were making the cut. Only when the elite schools are expensive do we accept them...
Very interesting review. I do want to take partial issue with this:
>There’s ample evidence that we can’t actually teach critical thinking divorced from content.
My issue with that statement is that people often incorrectly infer therefrom that students must master or memorize content before being asked to think critically. But that is not actually true; a student can "analyze Han and Roman attitudes toward technology" using the set of documents here: https://secure-media.collegeboard.org/apc/ap07_world_hist_frq.pdf -- including using evidence to support a thesis,; analyzing how the evidence might be incomplete or misleading; and what additional evidence might be useful to test the student's thesis.
Yeah, it shouldn’t be learn some content and *then* learn to critically think about it. Rather, it’s that you learn to think critically as a byproduct of learning some content. You have to be aiming at the content in order for the critical thinking to come.
It’s like how you clear your mind in yoga as a byproduct of paying attention to your breath, not by first paying attention to your breath and then clearing your mind.
Right, I should be more careful. It’s not a mere byproduct of learning content - it’s a byproduct of learning content in particular ways. But there has to be some content that you’re learning.
This review believes, but doesn't quite state explicitly: the purpose of school is to maximize motivation *for the high-structure learners*. This review seems to be telling us that we have to put everyone in the same classroom because this is the only way to get those high-structure learners to learn. I think the author would agree with a statement like: "we're going to sacrifice the learning of the top X% of the student body, because teaching them really slowly in the same classroom is the only way to keep the bottom kids learning at all."
The review at one point discusses something called "leveled reading", but dismisses it, because it "reduces the achievement of the readers who struggle the most". Great, but what does it do for the other readers? We never find out. (There's a link, but the linked article has no numbers in it.)
I can sort of understand this as a philosophy, but also I can completely understand why anyone with a smart kid would try to pull their kid out of it.
Because “high-structure learners” suck at learning, all schools must be made dumb and boring for everyone. No Child Left Behind because we are racing to the bottom.
For me school was demotivating so I'm not buying any of this. I learned a lot in this life but none of the things that stuck with me were learned in a classroom.
I enjoyed the essay, I agree with other commenters that the timing of this post works to its benefit.
I've noticed a number of comments so far suggesting different interventions to improve motivation. My read of the education literature (not an expert but did take a PhD field course in Economics of Education) is that basically almost every intervention under the sun has been tried and all of the ones that showed promise have disappointed once the programs have been scaled up. This includes interventions like paying students, paying parents, giving parents more information, behavioural nudges, etc.
I think one thing that tends to go underappreciated by those of us no longer in school is that genuine "learning" actually is fairly uncomfortable/hard. It's difficult to consistently stay outside of your comfort zone and encounter failure. Just think of how little "new" material the average adult actually learns once they're finished with their schooling.
On top of this I believe most adults have an inflated view of their own ability to learn. They equate learning some new procedure/process at their workplace to learning new content in school, but the two really aren't comparable. This has been made worse by graduate programs that are aimed at working professionals but still award a degree (i.e. MPP/MBA/etc.). These programs are often cash cows, they involve very little learning for most of the students. They tend to contain courses with 2nd year undergraduate course material, but the grading is done on the inflated graduate course curve. As a result, graduates end up getting a false view of what learning is really like.
This was a wonderful review, and as many commenters have said, a nice contrast to the Alpha School review.
A point I was waiting for that never came, was that the same student can be no structure, low structure or high structure in different domains. And that students can even move between categories over time.
Speaking for myself, I was solidly no structure at mathematics (I participated in the international math olympiad almost completely with self study), and high structure at writing - I just had no motivation. That was, until I took a philosophy of religion course in college and was able to write about things I actually cared about, and had opinions on. Suddenly I was no structure and able to ace writing assignments while actually enjoying writing.
And so, I'm unpersuaded by the assertion that a more personalized approach can't work for a typical low to high structure student. We haven't found an approach that consistently works, but that doesn't mean it doesn't exist.
This is close to my experience. I was good and self-taught at all the topics I was interested in: maths & sciences, really. I was bored in languages and literature… until I found some I was interested in (mostly philosophy and classical studies) and it turned out I was good at writing too.
This is what I find about getting philosophy grad students to learn math. A lot of them never thought they were good at math, but once they see that a philosophical question that they care about depends on something mathematical, they start to pay attention to that mathematical thing and start to improve.
>once they see that a philosophical question that they care about depends on something mathematical
I'm curious, what sort's of mathematical things do they find philosophical questions depend on? My knee-jerk guess is Gödel's incompleteness theorems and maybe Cantor's diagonalization argument??? Other examples?
My work is in formal epistemology, so I'm often showing that thinking about concepts like knowledge, and justification of belief, works better when you think about it in terms of probability. I personally was most interested in algebra and logic as an undergrad, and not interested in real analysis, but ended up teaching myself a lot more real analysis when I realized how central it was to thinking about conceptual issues for probability.
But you get other things in other areas of philosophy.
All of the classic diagonalization arguments (I count Cantor, Tarski, Gödel, and Turing as giving versions of it) end up being relevant for lots of thinking about meaning, mind, and language, as well as for understanding mathematics and computation. And in my department, a lot of people learn to code because they realize that making computational models of information flow in scientific communities is going to help them address the questions they're interested in for scientific epistemology.
Seems to me the real question is 'why did we spend so much time solving quadratic equations in school?' Really -- who cares what the solution to the equation of a parabola is? Are people continually coming up to you and saying, quick, quick, where is this parabola equation going to cross the y axis? Yeah, right, never. We spent so much time on it in school was because it was complicated enough that you had to spend some time learning it -- ie, we learnt it because it was a great way to use up time.
A thoughtful antidote to reform enthusiasts. Still I'm a bit puzzled. On the one hand schools exist to maximize motivation. But it seems that none of the three groups you've identified have their motivation enhanced. The unstructured are bored, the low structured develop a few habits but never catch fire; the high structured group loses steam the longer they stay in school. Many grow to hate school and learning. So whose motivation is augmented?
If I am told, "Do this unpleasant thing or I will cause you pain", I have not had my intrinsic motivation increased. But I have had my motivation to actually do it increase. He's talking about the second meaning of motivation.
I was surprised by the claim that deliberate practice doesn't work. Maybe we just aren't doing it right.
Sports are incredibly popular as a leisure activity. And during sports, even at an amateur level, you are generally being pushed to the edge of your abilities against people of a similar skill. You're constantly adapting, learning and then trying again with improved skill and knowledge. That's deliberate practice.
Several other popular leisure activities have a similar structure - chess and other board games, online pvp video games. Even solo activities like climbing or art where you are challenging yourself to do better rather than competing against another person.
Getting into a flow state can be very motivating. But it's tricky, it only works when you have incremental goals that are just within reach. If you push too far beyond that, then it becoming demotivating.
If I'm understanding the reviewer aright, I don't think they're saying that deliberate practice doesn't work — just that it's really hard to do it, b/c deliberate practice is hard and painful. (For what it's worth, K. A. Ericsson — the progenitor of "deliberate practice" as a concept — wrote in the book "Peak" that there's an only-slightly-inferior move he dubbed "purposeful practice" that is much easier to do in schools. Less autonomy needed.)
One minor correction - the text says that the high-structure students are the ones who are constantly in office hours. This is definitely wrong - the high-structure students are the ones who *should* be in office hours, but the fundamental problem is that there’s nothing that makes them come.
I have far more formal education than most barn cats, but a typical disdain for academic credentials. Beethoven never got an MFA. The only use I can discern for art school is so British musicians can drop out of it.
Still, how many of us insist on a self-taught brain surgeon? A Khan Academy trained aerospace engineer to design our airplanes? A really motivated amateur to operate a power grid?
Great article. I see that many commenters have said that giving more opportunities to high-performing students is not necessarily a problem. I enthusiastically agree.
In England, in 1944, the government set up a tripartite system where 11-year-olds took an IQ test, called the ‘11+’. The highest scoring 10%(ish) were sent to a grammar school which aimed to send those kids to university. The rest went to a Secondary Modern, which taught less academic subjects.
By the 1960s, it became obvious that this system was too successful, and society was becoming divided into middle-class kids who passed and working-class kids who didn’t. One unexpected problem was that the smarter working-class kids were passing, which left fewer smarter working-class kids to represent them in politics etc. The government shutdown this ‘selective’ system in the 1960s and replaced it with a ‘comprehensive” system.
Two counties stuck with the selective system and I lived in one. I passed the 11+ and went to a grammar school. I’ve had a very successful life. My brother failed the 11+, and went to work on a building site. He has had a very successful life too. It would have done him no good to stay in school for more years, learning calculus and Shakespeare.
I reviewed this before it was promoted, rating it low.
I have a variety of complaints, but the main issue I had was, I felt that the concept of "motivation" as pushed in this review wasn't actually well defined, and that this was used to push a class of Motte/Bailey. It starts by stating that schools aren't really about learning, and presents "motivation" as the new optimization target. It quickly asks "Motivation for What?" but the section doesn't actually address this question at all. It's roughly implied that it actually means "motivation to learn", which in turn maximizes learning.
Which ruins the entire thesis, because now he's back to arguing "best of all possible worlds" and "actually, this is the best way to accomplish learning". But, he doesn't present evidence for this case and actually, the evidence is starkly opposed to this case. You can't just handwave all of the evidence saying that traditional American schools are not very effective, and then conclude that traditional American schooling is the best possible education system.
Which is all to say, I don't think the author is actually being coherent, with the primary culprit being, replacement of standard terminology with non-standard terminology, then switching between the meaning of the standard terminology and non-standard terminology as necessary to win the argument.
"Age-graded" is mentioned a lot in this review. In the past, older students taught younger students in a "one room schoolhouse", with adult supervision. Maybe it doesn't scale, or doesn't develop motivation as well?
Oldest siblings have stronger personalities, are more of leaders (my observation). They practice being leaders with their younger siblings. They could (hopefully often do) care for their younger siblings as they lead them. They are sometimes a bit, or more than a bit, pathological. Optimistically, they just don't know better. We don't teach leadership and being responsible for other people in school (proportional to learning "material"). Youngest siblings don't practice leadership and care with younger siblings. Maybe it would be good to teach leadership and care, since many or most people end up being leaders to some extent, formally or informally. Having older students teach younger ones could help with that, perhaps. Also, perhaps teachers can delegate some of their workload to students. Students and teachers then would have more time for one-on-one tutoring for those who need it.
Great piece. While you allude briefly to this, I wanted to emphasize it more: students aren’t all one thing. I was a no-structure learner when it came to reading and writing. I could have learned those skills alone on a desert island. But I was a low-structure learner when it came to math. I needed somebody to walk me through it. So I needed school. I think a lot of students are like me; they might not need formal schooling for one area/topic, but they do for others. Some students need it for everything, but I think it is the rare, rare person who is no-structure across the board.
- when you keep nagging and threatening someone to study, because that is the only way to get information inside their head (and the person keeps fighting you all the time),
- when you help someone overcome distraction and procrastination, by giving them a time and place to learn (and the person may be annoyed for the first 5 minutes, but then enjoys it),
- when you give someone the right books and tell them the right words (and the person enjoys the help from the start),
We should distinguish between kids who actively resist learning, and kids who want to learn but need some kind of support. Both are "a structure", but of a different kind.
A pretty good review which rhymes very well with my own experience on the other side of the Atlantic.
I am starting to come up with a theory that normal schools actually are there to impart 'common knowledge', the knowledge that we know others know. Kind of like what you gain from an oracle saying at least one person have blue eyes.
The schools as we know was originally made so nations could organize their economies for war, which required some kind of common outlook on things.The countries that didn't got colonized by ones who did, and they then copied their own school systems.
I wonder what a world where we to loose this kind of common knowledge would look like. Right now everyone understands 'in democracy you have to hold fair elections', most people have heard of separation of powers even if they can't quite explain the concept, and some few people know about comparative advantage in production. Let us imagine this moved one step to the right, so most people had heard something about elections and only 10% knew courts are supposed to be independent of the normal government. Even if the smartest people were even smarter than they are now, I still think society in that world would be inherently unstable.
I think that this piece gets a lot of things very right about schooling. It's refreshingly clear and close to my own view. (Though it does in fact seem to be making the case that school is maximizing learning, when you take low- and high-structure children into account.)
For further evidence that age-graded schooling is doing something important in modern society, I'd also point to the lack of interesting variation among countries in terms of how they educate their kids. You don't really get the sense that there are competing systems out there.
I do think that there's a danger in slipping from "here is why things are the way they are" to "things HAVE to be the way they are." The piece makes some caveats in this direction, which I appreciated.
That said, I do find myself thinking that if AI tutors ever get as good as some people think they can be, we might figure out interesting ways to make it feel like "we're all doing the same thing" in the classroom while students are getting personalized tutoring. My students often love doing a bit of lightly-personalized practice with software (Deltamath). It's not that hard to imagine a better version of that software that teaches extension content during that block of time.
And at the college level, would it be that weird to have a "lecture" where the professor is setting the stage for a personalized lecture? Maybe you are assigned to a large biology lecture, no matter what bio class you're taking, so that the students present are taking many different courses. Class begins with a brief talk or exercise on a topic of broad interest, and then out come headphones and computers for some magic robot tutoring.
This is all sci-fi at the moment, and honestly I think it will be for the forseeable future, but I do wonder if there are other motivational structures that are possible.
>> "It's refreshingly clear and close to my own view."
Agreed! Michael, could you share how Deltamath "lightly personalizes" math practice?
>> "And at the college level, would it be that weird to have a "lecture" where the professor is setting the stage for a personalized lecture? ... Class begins with a brief talk or exercise on a topic of broad interest, and then out come headphones and computers for some magic robot tutoring."
I wonder if an even more motivating (just to stick to the theme o' the review) version of that might be that the professor opens up with a simple (but big, conceptually rich) question, then students go to their AIs for a fast, individualized dialogue to try to figure out the answer. (Something like this is what I do with my science classes for gifted kids, except minus the AI.)
I agree with the reviewer that if you define “personalized education” as completely self-directed learning then it will certainly be unsuccessful as a K-12 core academics ed model for almost every student. You cannot just stick some IXL loaded iPads in a room and hope the kids decide to work on math. But I see no reason why customized instruction, practice problems and testing for a specific student’s level must mean a self-directed learning environment with no rewards, accountability, structure, public visibility or recognition. This doesn’t have to look like someone signing up to take a MOOC during their spare time on a whim.
At one extreme, personalized ed may not be self-directed at all. It could be an academic tutor assigning you daily learning modules to work on, setting deadlines, handing out grades based on your performance relative to their goals, etc. The difference from regular school being that your assignments are customized to you - you have a tutor not a mass lecturer. The promise of today is that technology can enable a tutor working together with well-designed apps to coach a substantially higher number of students at the same time than would be possible in paper-and-pen days.
My impression is that Alpha is somewhere in the middle, allowing the students (perhaps with parents?) to set some learning speed goals and make some choices. But clearly there is a lot of accountability with all the testing. And they have a rewards systems they are developing with some controversy. It is interesting that the Alpha review independently came to the conclusion that their motivation system was a big part of what is driving success with that school model so far. I am sure much can be learned from sports, music and other non-academic activities about motivation models that don’t involve putting a bunch of kids of mixed ability in a room together to learn in lock-step. The Scouts use things like badges, for example.
One thing I didn’t understand from this review is how all the students being the same age was part of the motivation of the current school system. I understand the practical logistics behind this design, but how it relates to motivation was not clear. Is it more motivating for an 8th grader to sit completely lost in a math class watching same age peers easily answer the teacher’s questions than it is for the same 8th grader to sit beside a 6th grader who is equally unsure how to find an LCM?
My current beliefs after reading both recent entries on education:
1) I still believe that we could get much better at educating our top quintile of students.
2) I think economic growth and breakthrough innovation comes from the top tier of this top tier
3) I think the 80% could learn just as well or better and for less if we focused on getting them the basics and allowing them to shift to an apprenticeship type model where desired.
4) I think AI coaches supplemented with group activities and human teachers will lead to amazing breakthroughs in the near future (for self motivated learners)
5) I am extremely skeptical about the ability of the current education establishment to redesign how education can or should work. They and their way of thinking are more of a hindrance, on net, than a help.
I am probably wrong on one or more of these though.
I went to a selective high school that had internal tracking in every mandatory subject: math, science, English, and social studies all had advanced and regular classes. Students who were in one advanced class were typically in advanced classes for all those subjects, including me. One year I was dating someone in a regular English class and attended it with him during a free period (as one does). I was surprised to see that there was daily homework, about a paragraph of writing per night, plus regular quizzes on the content of the readings. The class didn't seem to be doing well in either regard. I was also a bit surprised by the silence of the class, and how happy (or at least accepting) the teacher seemed to be that I was participating in the class despite not actually being enrolled.
In advanced English classes we wrote 1-2 essays per book (with no homework most days besides reading and having thoughts on the reading) and the conversation never stopped and infrequently went off-topic. Occasionally we finished books ahead of time because everyone had read ahead. Similarly, in advanced science classes, students often read unassigned material and asked teachers for additional conceptual guidance, sometimes after class. (Teachers also seemed to be nicer, and more interested in helping interested students, in the advanced track. Regular track teachers were more likely to have reputations for being mean.) None of this stuff was graded, whereas each daily bit of homework in regular classes seemed to be an effort to tip students over the passing line.
It's clear in that the advanced track served students who needed less structure at least in those subjects (if not overall), but it also seemed that the regular track served to whitewash the performance of lower-performing students. "Look at these 18 paragraphs they've written this semester", you could say, "they've done 65% of the work." And maybe that adds up to 65% understanding of the books read or narrative devices discussed or self-expression opportunities afforded.
As a 'low-structure learner', I've found that one of the reasons motivation is hard without structure is that plenty of entities are willing to make you spend a lot of time on bullshit – if they aren't paying anything for a teacher. I've taken stupid classes, but I've done a lot more stupid online modules. Teachers are expensive, and that's a honest signal that someone has at least some real investment in teaching you something worth learning.
The most promising alternative to public school that I've seen is a thing called modular learning -- at least the version developed and practiced by Jeremy Howard (tech genius -- https://en.wikipedia.org/wiki/Jeremy_Howard_(entrepreneur)) and a woman named Manisha Snoyer. It's homeschooling done in a way with some structure, curated materials, and some shared online classes and learning experiences with other families who are taking the approach.
I'm not suggesting it as a solution for the problem of how to educate all kids -- though who knows, it might work -- but it does seem to me like a flexible approach that could be adapted to work for many kids.
School isn't quite like Chesterton's fence where it's been around so long we forgot why it was created. Instead (as you note paragraphs after invoking the fence), mass education is relatively recent. The reason people are often impressed by old timey written tests is because a smaller fraction of the population was taking such tests.
IQ is increasingly heritable with age. Just as school effects fade out during summer vacation, so do such effects when students eventually graduate from school. Thus gains from school are temporary. Per James Flynn, gains in IQ over time don't appear to be gains in g either, so it can be difficult to generalize effects of such gains from other effects of IQ.
> Classic studies like Asch’s line experiments, where 75% of participants denied obvious truths to match group answers
75% did that at least once, but that's out of 12 incorrect answers from a larger set of 18 and "The average participant gave in to the group three times out of 12" https://www.overcomingbias.com/p/conformity-mythhtml
> Schools spend a huge fraction of their resources on special education, providing the structure and systems that those students need.
Why? They aren't a huge fraction of the students, and if we think of spending as an investment, there isn't that much of a return for it. The focus should be on cutting costs since there is such little benefit. Similarly, the most disruptive students might be abandoned for the benefit of the rest https://www.overcomingbias.com/p/helpful-inequalityhtml
> Maybe the best way education can contribute to all of these causes is to focus on providing a broad, basic education to as many children as possible, and maybe our current system is the best idea we’ve had so far for doing so.
What is the reason for believing either of those things? As you acknowledge, the later purposes were added after the system was first created for the earliest purpose, and it would be astonishingly lucky if it were the case that it just happened to serve those later purposes it wasn't originally designed for. Recall also that the original political purpose of shoring up democracy preceded the rise of schools of education and critical theory such that my mom was required to study (distillations of) Pedagogy of the Oppressed to teach Special Ed. There isn't just conservation in the school system, there is also rot https://www.overcomingbias.com/p/what-makes-stuff-rothtml
My problem with this review, which could well be right in some respects, is that there is a lot of speculation without appropriate empirical evidence. For starters, much of the evidence cited on the effects of tracking does not even measure tracking in the sense that many ACX readers will be thinking of and the few I know of that do tell a slightly different story.
If you open the linked meta-analysis and look at some of the individual studies, you will see that the studies on overall achievement typically look at the effects of separating students at the high school level into different schools where they often learn totally different curriculums, relative to comprehensive schools. However, those comprehensive schools track their students according to ability internally. For example, every UK comprehensive high school (11-18 for foreign readers) continues to group their students by ability, but a study like Hanushek and Woessman (2006) is treating us as stopping tracking when we move from grammar schools to comprehensive. All that really happened is that we abolished the substandard secondary modern vocational schools. I never shared a Maths, English, or Science classroom with anyone in the bottom two thirds of the ability distribution during high school, but all these studies will say that I wasn't tracked. Theoretically, the move to comprehensives could improve the ability matches between students and their classroom peers. If some students are late developers or richer parents game the admissions system, you could have bottom half ability students stuck in the higher school who get left behind and top half ability students stuck in the lower school with no opportunity to progress.
Meanwhile, the randomised trials that I am aware of that actually measure tailoring the classroom according to ability show very different results, with the caveat that they are all from developing nations. One set of evidence comes from the teaching at the right level randomised trials in India (link: https://www.nber.org/papers/w22746). The cleanest example here comes from Haryana where they grouped students together by ability across year groups for an hour a day for a whole year for Hindi lessons and students progressed by 0.15 standard deviations relative to control. That is likely close to a whole year's typical learning (source: https://blogs.worldbank.org/en/impactevaluations/simpler-way-communicate-learning-gains). The Uttar Pradesh intervention did make use of outside volunteers in their teaching alongside ability grouping, but the effects were four times as large! There have also been computer-assisted interventions for Maths that target students by ability level to substantial effect (https://academic.oup.com/qje/article/122/3/1235/1879525). The other example comes from Kenya where Duflo, Dupas, and Kremer randomly assigned schools to track students by ability for 18 months and found that learning improved by around 0.2 standard deviations (link: https://www.aeaweb.org/articles?id=10.1257/aer.101.5.1739) - about a year's extra learning again. This study, although likely not subject to publication bias, is a bit weaker as it's merely p<0.05 rather than p<<0.01 like the Indian interventions.
I'd also add that I think there are other areas that I think are poorly evidenced or not evidenced. The three things that stand out are the claims that education quality doesn't matter much, learning loss wasn't that heavily driven by length of school closures, and that you can't group by ability rather than age group. The first isn't true between this study of the UK that also outlines 4 US studies in Table 1 (https://ifs.org.uk/sites/default/files/2023-09/WP202324-Exploiting-discontinuities-in-secondary-school-attendance-to-evaluate-value-added_0.pdf), there are meaningful causal differences in quality between schools. Additionally, the school vouchers study mentioned in the deBoer package is about sending poor students to wealthy private school going badly, which could very well be due to them being low ability students there who can't keep up with their peers. I quibble the learning loss point because the length of school closures was enough to fully explain learning losses in the one study I have seen (https://www.aeaweb.org/articles?id=10.1257/aeri.20210748). My uncertainty on the unevidenced age group part is that i) grouping by ability rather than age did work in Haryana and ii) you can also spin arguments about why doing so may improve student effort. Maybe students falling behind would work harder to keep up with their friends who are moving up, or low ability students would stop acting out in an effort to show off if the only people they can show off to are several years younger. I don't know because it's almost never tested.
Finally, I'd also throw significant doubt on the likelihood of the linked meta study to even measure the thing it is measuring accurately when it cites Manning and Pischke (2006) as evidence of the effects of tracking (in the separate schools, not classes sense) on education inequalities when just reading the abstract of their paper would tell you that their argument is that you cannot reliably estimate any causal effects using the available data.
Sorry if there are any sloppy mistakes or omissions. I've only spent about an hour on this post and have relied on studies I was already familiar with in my work as a labour and education economist and although I have read all of the studies at some point, not all of them were read forensically. I also have not proof read this, so some parts may be unclear.
What a thoughtful pushback to the too-cheery hype of educational reform! (And I say this as an educational reformer.)
>> "Schools... are designed to maximize motivation, but motivation is hard to maximize, and schools don’t do a great job."
I really appreciate the reviewer's identification of motivation as the bottleneck. Indeed, if the problem were anything else, then the spread of libraries should have led to the messianic age of education!
The reason I think massive improvement in education is possible is that there's a reservoir of motivation already inside schools that has not yet been systematically tapped: the content of the curriculum.
Math, science, history, and so on — down to spelling and, I don't know, small engine repair — are made up of innumerable small bits that have fascinated people in the past (that's how they got in the curriculum). The trouble is that they're pressed flat and delivered as unpalatable by the time they get to the classroom. The revivification of academic content is the best hope we have for an educational revolution that scales. (As a plus, it also doesn't require we figure out how to revolutionize a system that's been impervious to 100 years of reform. In any case, this is what I was trying to get at with my ACX book review in 2023.)
The reviewer comes across as a wise and careful thinker. If they're reading, I'd love to chat — feel free to reach out to me! (I won't compromise your anonymity.)
Who exactly was fascinated by those things in the past? Wasn't it a highly selected group, for any given topic? So this doesn't seem to tell us much about whether we can get the general population to engage with them.
Sharp point! But it also works the other way: we can look to see what aspects of the topic motivated them, so we're not working from zero. If this seems abstract, feel free to give an example of something in the K–12 curriculum that most people find dull, and I'll see if I can show how it can be re-humanized. (This is what I do for a living...)
I dispute your claim that schools are successfully bringing knowledge to even the low self-motivation students, because they often don't work to even teach in the first place. It's not just the "learning loss" like what Scott talks about where he forgot pretty much everything he doesn't use or see reinforced in pop culture; I've tutored multiple students in calculus classes who can't find the area of a rectangle. If their evaluations were working at all they wouldn't have been advanced to this class in the first place.
How is this possible? Imagine a teacher fails to teach anyone anything at all (standardized tests and personal experience both confirm this can happen). Students who fail would be held back a year, but holding back children would anger parents a lot, so they are sent to the next year anyway with a fake grade. Obviously, children are considered people with agency, so even if they are able to learn the subject elsewhere on their own they are still sent into the class, by the police if necessary, even if their teacher does nothing but scream into their faces and get really angry at people who ask clarifying questions, like Ms. Mandeville. (The actual name of one of my teachers!)
Most of the time schools aren't that obviously bad, I admit. Possibly most schools "merely" teach things that end up forgotten while making kids hate learning enough that they stop being curious about the topic if they started out curious. But the school system is so dysfunctional that it can't even detect if anyone learned anything at all.
This review rubs me the wrong way quite a lot; unfortunately, I can't give it the time it would deserve, so here is a short version. I will start with the parts that I agree with, to get it out of the way.
1) I fully agree that you can't "teach critical thinking" divorced from content. It is one of those things that sound great but when you look for substance behind the words, you won't find any. (Called "applause lights" in the Sequences.) How could someone think critically about e.g. World War 2, if that person had no idea when the war happened, who participated in it, and other similar encyclopedical knowledge. What would their opinions be based on, if not facts? A random edgy tweet, or a convincing (for someone who doesn't know any of the facts) article on a blog?
It actually works the other way round. You get some data in your head, and sometimes they connect, maybe in a way you did not expect. (And you cannot simply look them up on Google, because you do not know what will connect to what.)
I would even go further and say that, as far as I know, we do not have any evidence that "critical thinking" can be taught *at all*. I think there was an experiment with some university students who studied critical thinking, which showed that after passing the exams they were not any better at critical thinking compared to their classmates who didn't attend those lessons. (Also, "knowing about biases can hurt people" in the Sequences; we all probably know people who only learned the techniques of "critical thinking" superficially so that they can use them to reject inconvenient arguments.)
2) I agree about the self-reinforcing nature of motivation + study habits + skills/knowledge. (But I disagree that the third one is "intelligence".) Even getting slightly ahead of your peers can start the feedback loop where people praise you for your skills, you feel good about the topic in general, which makes you more willing to study it further, which further increases your skills, etc. And there is an equivalent opposite loop where falling slightly behind your peers makes it all feel hopeless, because you would have to work hard only to become average, which is no big success.
But from my perspective, this makes separating children at school by their physical age a horribly bad idea. That's how you create the losers! I mean, if someone is e.g. literally retarded, that is going to be obvious no matter what. But if someone is merely a few months behind their peers, the school system makes it super visible by contrasting them with their peers every day and giving them bad grades. (To compare with the life of adults, no one cares if you e.g. get your driving license one year later.)
3) Yes, it is easy to create an awesome school by having a random idea (anything that is not actively very harmful) and selecting only for good students. Good students create great learning environment, and you can then attribute that to your excellent idea, or your excellent teachers, or whatever. This can create a self-sustaining loop, when the school becomes famous, so many students apply, so the school can keep choosing the best ones.
This shouldn't be taken as a proof that good ideas are impossible. But it invites a lot of skepticism about ideas even after you have seen a successful school that follows them: maybe the true success is in the constellation of the students, not in the idea.
...okay, and now the remainign 90% that I disagree about...
For starters, I roll to disbelieve in the studies on increasing IQ. That doesn't even pass the smell test: IQ is defined *relatively* to people of the same age, with the average being 100. So if the school attendance is mandatory, and the average of the kids in the first grade is IQ 100, and the average of the kids in the second grade is IQ 100, and so on... where the hell is that 1-5 IQ points per year increase? Does not make sense, mathematically.
(Or do you mean that the kids in the second grade are smarter relatively to the kids in the first grade? Yeah, that's called "growing up"; I am sure it happens a lot outside the school, too.)
The repeated topic in this review is that almost all kids need some kind of external support / motivation / pressure / direction / structure / whatever we may choose to call it; and that although some kids only need a little or none, most kids need a lot. Without the system, the kids who need a lot of the external force would utterly fail. And... that's true... but it also seems irrelevant when comparing the existing school system to the alternatives, except for those that advocate zero use of external force, such as unschooling, some specifically American misinterpretations of constructivism, and the naive idea that if you give computer lessons to the kids you will never need to check whether they have actually turned on the computer and actually clicked on the lessons. But those are far from the only possible alternatives.
The original experiments with MOOCs failed, as far as I know, mostly because they were done in a way "here is the digital content, make sure to finish it before the end of the semester", and most students ignored it until one week before the end of the semester... and then it was too late to learn in before the exams. Yes, that specific approach to using MOOCs fails. But instead of rejecting the digital education completely, I think the proper lesson is to make sure that the students keep doing the lessons during the semester. Which, once you think about it, is trivial to implement: at the end of each week, check how many % of the semester have passed, how many % of content each student completed, send them an e-mail if they are behind, and send an e-mail also to their teacher is they are significantly behind.
These days, many MOOCs require you to proceed at a predetermined speed, which is horrible for an adult person who is not a student, doesn't have a predictable schedule, and just wanted to study something in their free time out of curiosity... but it addresses the problem for the students who use those MOOCs at school. There are often also forums where the virtual classmates can discuss the lessons and ask each other questions.
Similarly, the main problem with COVID-era remote education, in my opinion, was that people were totally unprepared for it. You had an improvized Zoom call where the kids didn't even hear the lesson properly, and... often, that was all. Yep, didn't work. But if you knew e.g. one year in advance that you will have to spend a few months teaching remotely, there is so much you could do better. For starters, you could prepare sheets with exercises, and deliver them by mail to students' homes. (Or, depending on the socioeconomic class, just send them some PDFs and tell them to print them.) You could prepare a nice video of the explanation, with clear image and sound, and upload it to YouTube. (If multiple teachers did this, you could also send your students links to other teachers explaining the same topic, so maybe they would benefit from having the same thing explained in several ways.) But more importantly, this could all be connected: watch a video... keep answering related questions on the paper... communicate in chat... and then do an automated exam. Maybe this wouldn't work perfectly, but I think it would work much better than mere improvised Zoom calls.
"In general, research has found that leveled reading reduces the achievement of the readers who struggle the most. We might intuitively think that reading an easier book would benefit students who have weaker reading skills, but that intuition seems to be wrong."
My intuition would suggest to give everyone an easier text first, more complicated text later, after they can read the easier text correctly. Where can I go collect my Nobel prize for education?
As a sidenote, giving up at the first failure seems like a repeated pattern in debates about education. Imagine someone talking about e.g. cooking the same way: "Once I tried to cook a soup. It was too salty. The idea of 'cooking a soup' is debunked. Anyone who suggests that you can cook a soup at home is clearly wrong, they should read my report instead." I find it ironic that the school system is so bad at learning. Then again, school has practically redefined "learning" to mean "memorizing the words of the authority", so perhaps the idea of learning by experimenting is at a blind spot of most experts on the school system.
Another sidenote, even the words "no-structure", "low-structure", "high-structure" rub me the wrong way, because they are so... structure-centric. It's like calling a bird outside the cage a "no-cage bird", which kinda suggests that the cage is the bird's natural habitat, and it's the absence of the cage that makes this specific bird weird. I would prefer an opposite perspective, e.g. to call the students high-agency, low-agency, and no-agency... or some word other than agency, but the focus should be on the quality that makes learning happen (and which, when low or missing, can be substituted by the structure).
"incremental rehearsal is a highly-structured way to teach students multiplication facts."
Well, that kinda sounds like something a computer could help with, just saying.
"The no-structure learners will always be bored, as long as we are committed to putting them into classrooms where everyone learns the same thing."
Wrong. There is a vast difference between "learning the same thing" and "learning the thing that you already know... at an excruciatingly slow speed... every day... for years". Both can be boring, but it's not the same quality; given the choice, I would choose the first one in a heartbeat.
Also, maybe we shouldn't teach the boring things, or we shouldn't teach them the boring way? Why are we even doing that? Yeah, because that's the only way that works well with the high-structure students. Basically, the students who resist learning require an approach that is almost the *opposite* of someone who wants to learn. It's like, for the student who wants to learn, you build the track; for the student who doesn't want to learn, you build the barriers along the track so that he can't escape.
"[Kids at gifted school] would be doing more advanced work than peers in regular schools, absolutely. But often only by a year or two."
Well, a year or two is a lot. Imagine coming to the university two years earlier. For example, if a person wants to spend a fixed number of years at school, how much more they could learn if they could replace two years at elementary school listening to the stuff they already know, with two extra years of university education? The same cost for the educational system; much better outcome.
From my perspective, the problem with many gifted schools is that they focus too much on intelligence, and fail to notice things such as love of learning. If you combine intelligence with love of learning, you get magic. If you combine intelligence with the lack of love of learning, you get bored students sitting in the class for gifted kids.
Note that the love of learning is *not* the same as "no-structure". Those are two different dimensions. If you make a voluntary class, the people who will come are those who love to learn, and prefer to have some structure for that. I know many people like that.
"The core challenge of compulsory public education is motivation. The best solution we’ve found is to send kids to school beginning at age 5 (or earlier if we can), before they can reliably form long-term episodic memories."
Are you insane?! School is typically the thing that *kills* motivation in children. Small children are curious about everything, they can't stop asking question. It is school that makes them associate learning with boredom. It is school that associates learning with punishment (even if merely verbal and social).
"The system motivates high-structure learners to keep up with their peers, though that motivation does gradually fade over time."
So we basically sacrifice motivation of some kids to create motivation for other kids... and even that doesn't work well in long term. Cool!
"Maybe the best way education can contribute to all of these causes is to focus on providing a broad, basic education to as many children as possible, and maybe our current system is the best idea we’ve had so far for doing so."
Maybe the current system is the best for those children who don't have a hope to achieve anything beyond the basic education. And maybe those children are a majority of the population, like maybe even 80% of it, although I doubt it.
It would still make sense to have a separate track for the remaining 20%. And high school is too late.
As a former teacher, it was my experience that the class is often slowed down by one or two kids who are either clearly uncooperative, or somehow damaged. Maybe those schools were not representative for the population at large, and the actual ratio is different at other places; I don't know. Sometimes, if such child became sick for a few weeks, the classroom experience changed dramatically in their absence... and then immediately reverted back. As a result, sometimes 18 out of 20 students cannot get the education they might get. And this is considered perfectly acceptable. It is democratic, inclusive, insert your favorite buzzword... Yes, you can get a dramatically better school by simply *not* admitting kids like this, and then people complain that this is cheating. From my perspective, it is sanity. Instead of complaining that it is unfair when only a few schools do it, all schools should be doing it. (That is, except for a few selected schools that would be specialized for kids like this.)
Basically, we are trying to make the school same for everyone, and then it barely serves the lowest common denominator. All improvements will be criticized that they only work for a certain subset of students. But from my perspective, that's okay. The students who can get a better system, should get a better system; different systems for different students. And the students who couldn't work in a system different from the current one, should stay in the current one. But they should not be allowed to drag everyone else down.
Without making any claims about whether any particular teaching technique works, it seems pretty clear to me that "critical thinking divorced from content" is a coherent concept and in principle is learnable. Our esteemed host made the case for it better than I could, nearly a decade ago (https://slatestarcodex.com/2015/08/04/contra-hallquist-on-scientific-rationality/):
"If we want to get all hypothetical, we can imagine some kind of theorizing contest between a totally irrational person with an encyclopaedic knowledge of cell biology, versus a very rational person who knows nothing at all about the subject. Who would win? Well, who cares? Whoever wins, *we lose*. We lose because I want the people working on curing cancer to be good at both cell biology *and* thinking clearly, to know both the parts of science specific to their own field and the parts of science that are as broad as Thought itself. I have *seen* what happens when people know everything about cell biology and nothing about rationality. You get AMRI Nutrigenomics [1], where a bunch of people with PhDs and MDs give a breathtakingly beautiful analysis of the complexities of the methylation cycle, then use it to prove that vaccines cause autism. By all means, know as much about methylation as they do! But you’ve also got to have something they’re missing!"
This only addresses one piece of the problem but -- what if in order to vote people had to demonstrate decent knowledge of structure of government, the last 100 years of history, laws, political parties, taxes, treaties, voting, and let's say 3 current issues? AI can generate lots of tests and ensure that they are of equivalent difficulty, and score them, both multiple choice items and short answer.
Oh sorry I thought that every time the word "school" appeared in the review the writer meant "Government." Like that "school" was just a slang term. I misunderstood the whole review! Oh dear I am such a silly.
Is your point that if we disenfranchised the ignorant then we could just give up on educating unmotivated students, because their ignorance can't hurt their fellow citizens? I do not think that that follows. In a highly complex information-age society, externalities are everywhere, not just in electoral politics. You want the people you transact with in the marketplace, or who live next door to you or pass you on the street, to have a baseline understanding of how the world works and of our shared culture; it doesn't always matter, but sometimes it does. And being uneducated puts you yourself at a disadvantage as well, as it makes it harder to tell when you're being duped. Yes, there's a lot of waste and uselessness in the K-12 curriculum, both for unmotivated students and for motivated ones, but this does not move me to give up entirely on the concept of a universal baseline education, albeit maybe a smaller one than we presently have.
Also, disenfranchising a large fraction of the population would mean that elected officials would be free to entirely disregard their interests, and that seems *extremely* bad and I would have to see a damned strong case that it was worth the cost.
Importantly DK is assuming an ultimate point to school, inside the perceptive point that it's about "motivation." Motivation for what; learning; this elides the point.
Consider different functions of schools:
1) Teach kids useful information that will make them more effective and productive adults.
2) Day care. Get kids out of the way so the adults can work.
2a) Divert kids away from worse activities like watching TV all day, drug use, unprotected sex, crime.
3) Produce a ranking of the kids relative to each other.
4) Stealth social services. A square meal a day, and put kids in contact with some caring adults.
5) Produce entertainment: school plays, band, quiz bowls, beauty pageants, sporting events.
6) Wealth transfer from prosperous blue areas to unproductive red areas via teacher, administrator and staff salaries and spending on school infrastructure.
7) Political indoctrination, e.g. the Revolutionary War is an origin story for the US meant to create political cohesion.
I agree with DK's cogent point well-made that school is structured for motivation across the whole population. But the "motivation" is for "perpetuation of school". Schools have evolved to perpetuate schools, unsurprisingly.
Upsettingly for people who like to envision better schools, not going to school turns out to be quite bad. Smartphones in schools are undoing a chief purpose of modern schools, to stop kids from watching TV all day. Smartphones have put TV right into schools. Movements to ban phones in school are very good.
The Federal Reserve, of all people, did a careful study of how kids spend their non-school time, and found that diverting kids from bad activities like watching TV was important:
The idea of “high structure kids” and “low structure kids” is important to a lot of the author's argument. How established and validated are these categories?
There are some dimensions on which almost everybody falls into one of a few categories: gender, blood type, eye color, handedness. But on most dimensions people are distributed in some approximation of a bell curve: most people are in the middle, and a considerably smaller number are quite a bit above or below the middle range. Some in the absence of evidence to the contrary, I'd expect that most kids are medium-structure.
And then, on the dimensions on which variance takes the bell shape, a lot of them turn out to be situation-specific. For instance, consider aggressiveness. If you look at most people across a variety of situations, you will see a lot of variation in how aggressive they are. They might be very aggressive chess players, driving grimly towards the endgame, but very casual when playing tennis, chatting about this and that while they play. They might push very hard in work discussions when certain issues come up that they care a lot about, but shrug and say “whatever” about other issues. They might be very irritable with other drivers, but patient and gentle with their teenage daughter even when she’s being quite difficult. In fact there are not many “qualities” at all on which people's scores are fairly stable across different situations. Intelligence is pretty special because it is one of those qualities. There is a lot less variation across situations: People able to remember and recite a long string of digits also tend to have a llarge vocabularies, and to be good at mental math and at reasoning, and to have la large store of general knowledge. So in the absence of evidence that how structure-dependent somebody is stays the same across settings and situations, I'd expect that most kids vary quite a lot.
So unless there’s some pretty solid evidence that most kids are either high structure or low structure, and that they stay in whatever category they’re in across a lot of situations, then a lot of the author's train of thought about schooling and what works is unfounded. Is there in fact some well-validated psychometric measure of high/low structure, one that shows the distribution tends towards being bimodal and that kids’ high- or low-structure membership tends to stay the same across school subjects?
I’m told one-room schools were even better than grouping by age, because you could hear the lessons from higher grades before you got there. My father’s family was from a tiny rural fishing village. When he moved to California at 16, they were years ahead of their piers in education.
Good morning, I made a manifold market for this years contest!
https://manifold.markets/BayesianTom/who-will-win-acxs-everythingexceptb
Will "other" turn into specific reviews as time goes on, or are the choices fixed?
I will add new choices every friday. I didnt want to add all options beforehand because the roster isnt fixed yet (for example, we got this honorable review turned into a contender this week).
If you want to bet that the winner will be neither of the current choices you can bet on "Other" and your bet will be converted to any new choice added in the future automatically
Gotcha. My intuition says that the odds for the first two reviews will be a lot higher now, just by lack of specific other choices at the moment. I've read all the finalists already, and I still think AlphaSchool will win, but I think I can get better odds (in the sense of lower probability and therefore more margin) later on.
Are all the finalists listed somewhere? I've probably read all the finalists, because I read most of the review submissions (and almost all the reviews which were not of works of fiction,) but I don't know which made it to the finals apart from the two which have been posted so far.
For what it's worth, I had one specific frontrunner among my votes, and it wasn't either of the ones posted so far.
https://www.astralcodexten.com/p/open-thread-387, point 3.
Thanks, I missed that last time.
It looks like all my top-scoring reviews made it onto the list, but there are a few there which surprise me, which made it into the finals or honorable mentions despite my giving them unexceptional scores.
I don’t think schools are that good at creating strong motivation but it’s true that they provide an overall structure that semi-forces many kids to learn something. However I think we’re at the point that we can offer the equivalent of one-on-one tutoring for every student. “Unstructured learners” can learn directly from books, but better personalized learning along with a supporting social structure can help everyone.
The issue will be how the personalised learning forces kids to complete as much as they can (instead of confusing lack of interest with being stupid and not making them do much). "Motivation" is being used in slightly the wrong way in this essay; many (most?) children don't want to learn whatever the school's trying to teach them, and if given the opportunity to slack off they will (these are the "low structure" kids); school motivates them by forcing them to sit in a classroom and not focus on anything else, but it doesn't make them more motivated in general. With personalised learning, they'll learn at a snail's pace because they don't care about advancing through the material. Or the personalised system will declare that they should be working faster based on [brain scans/genome testing/extrapolation from some kind of reasoning test] and force them to learn at a somewhat arbitrarily faster pace, in which case they may as well be in a classroom.
I hope that personalized learning will not only mean different speed (although that alone is already a huge improvement over status quo), but also different choice of topics.
What I would like to see is some core "everyone needs to know this", followed by tons of optional topics. So that for an average child, 50% of school would be optional topics, and for a gifted child maybe even 90%. (Every year, you would have to gain at least X core credits, and at least Y total credits.) You should totally have specialized topics, such as "classification of dinosaurs".
The advantage of computer teaching is that those topics could *accumulate* over time: maybe the first year there wouldn't be much of a choice, but 10 years later you could have tons of optional topics. (The system should be open, something like SCORM.)
Motivation is the parent's job, not the school's. If a kid doesn't want to learn then he's probably gonna have a bad life. You'll do more net harm wasting bureaucratic resources trying to reach that kid than you will by shoving him out of the way so the smart kids can learn.
You're talking about the vast majority here, would you prefer them to get no education at all, the whole school system dedicated to the 5%? Can't we do it the other way around: a school system for the majority and one-on-one tutoring or online learning for the motivated 5%?
The vast majority already don't get any education at all. I say we stop pretending that they are. Total government spending on education is in the neighborhood of a trillion dollars per year, and most of that is on maintaining the kabuki theater that everyone can learn. Just accept that it's not reasonable to have our current education goals. Optimize for the top 20% because that's who makes the economy go. The rest can be tracked into low-cost high-efficiency warehouses where modest but achievable goals can be hit: learn to read, basic arithmetic, some cultural indoctrination, and maybe teach a trade. Everything else is just a waste of taxpayer dollars in my view.
You are almost entirely wrong that the top 20% is what makes the economy go. Our economy is built on the backs of relatively cheap educated workers. Schools do such a good job of this that these people are invisible to you. But read back to the post-war period when all this stuff was booming -- it wasn't being taken for granted then.
That may have been true in 1950, but in my view part of the reason for the GDP/median wage decoupling in the 70's is that it's increasingly not true. We have an increasingly complex economy and fewer and fewer people can meaningfully contribute to it. I actually think that 20% figure is generous: it's the entrepreneurs and innovators in the top 5% (or 1%) who really matter and that is reflected in the increased share of total wealth that they now hold relative to 1970. The "backs of relatively cheap educated workers" are actually the other 15% of the top 20. Below the top 20% or so and you're not actually dealing with knowledge workers. They may have a mediocre college degree but they're essentially just moving boxes around, answering telephones, filing paperwork, etc. They actually don't need the education, it's simply an aspirational status marker. That's why real salaries for non-STEM degree holders haven't really moved in 40 years.
I think the 20% number is being generous here. Economic growth comes from innovation and technology driven by the top of the top and (to Michael's point) implemented by the majority who need to be able to read, write, do sums and carry out the countless manual and interactive tasks with a smile.
Agreed. My point is that the white-collar hoi polloi below the top 20% don't need any real educational training apart from basic literacy and numeracy, and that can be attained much more efficiently than it is now.
Minor editorial comment: "low-structure learners" is referenced once before it is defined.
I was tripped up by that too in an otherwise very polished review. After learning the term, I decided maybe in some contexts I'm a low-structure learner, such as this book review contest. I hadn't read this review earlier; in fact I hadn't even noticed it; in fact I read only a handful of the reviews; in fact I could scarcely get through glancing at the titles of the reviews because the structure of the Google doc was a bit too low for me. Now I've read it precisely because it was presented for all to read, today's assignment. I don't have much to say on the subject, but will get enjoyment out of the responses others in "the class" share.
Interestingly, school itself did not function this way for me.
Freddie, if you're out there please respond
I would like to boost the deBoer bat signal!
I think this was great read after the previous review. I don't know if the order was intentional or not, but the contrast makes it a perfect companion.
> You might think that we’ve found the solution to tracking. We just need to get all the no- and low-structure learners together and let them move much faster. Here’s the issue. The no-structure learners will always be bored, as long as we are committed to putting them into classrooms where everyone learns the same thing. And those classrooms where everyone learns the same thing are exactly what the low-structure learners need. As soon as you create a higher track there will be a ton of demand for it. Parents will insist that their kid join. And as it grows, it won’t be able to accelerate very quickly. You still need the structure of a classroom where everyone is learning the same thing, and that just isn’t a very efficient way to teach.
I think this might be the fate of every innovative school that tries to serve the unstructured learners. As a school suited for the top 5% of unstructured learners starts attracting a more general student body, largely due to its success at producing high-performing students, it will be forced to adopt more generalized structured learning to accommodate for those new students. Administrators will justify every step as simply adjusting their teaching style and structure to work for larger groups, and at the end of the day you end up with a different, but not revolutionarily different structure to what was already available in the school market.
Maybe the only solution is an eternal pattern of innovation --> decay --> innovation. At least it gives us something interesting and difficult to worry about! Or maybe AI will genuinely be transformative, allowing for extremely tailored learning that identifies what style works for each student, and gives them more of that. Such a thing would only be possible by throwing large amounts of resources at the problem, which is impracticable now.
Anecdotally, Montessori motivated my children much better than public schools have. In Montessori, in particular, children don't all learn the same things. They are encouraged to work on the things that they want to. All children are taught the same works, but what children work on are largely self-motivated. This worked quite well for us, and also worked very well for the parents that I speak to for their children.
It would be valuable if you could describe in more detail how the Montessori system works (in practice, that is). My first question is what is the student/teacher ratio? It seems intuitively plausible that if a student gets plenty of attention from a teacher constantly assessing what the student is motivated about right now, and carefully feeds them all of the curriculum over time, that would work well. But it wouldn't scale.
There’s a lot of information online and I’m not an expert. (Also any school can call itself Montessori so likely a wide range in what schools actually do). Relatively large classes, with (usually) one teacher and often one assistant, student ages span three grades. The room is full of different ‘work’, and the children are allowed to choose what they do but only after they have been trained how to use/do a particular piece of work. Each piece of work often has different levels, which the kids progress through. (Search Montessori beads for an example of math work). Older children often teach younger children how to use certain pieces of work. I think lots of folks think that Montessori students have more freedom than they do — it’s a highly structured environment that nonetheless allows the children to make decisions for themselves within that structure. Lots of emphasis on using and putting away the work in a prescribed way. Like most private schools it ends up being very selective in who attends and who teaches. I also don’t know anything about how it operates past third grade (US). Most Montessori schools stop at third grade or earlier. My sense is that it works well for no, low, and high structure students described in this review because it is in a classroom with lots of structure.
> any school can call itself Montessori
There is such thing as certified Montessori school, but most "Montessori schools" out there are not.
> Older children often teach younger children how to use certain pieces of work.
This is one of the crucial components missing from traditional schools. Kids can teach each other... that is, if you allow them. Some older kids will be happy to gain status by teaching the younger ones, and many younger ones will be happy to get attention from the older ones.
Traditional school bans this, and then people complain about the ratio of teachers to students. Well, it's the school system that made the ratio artificially low in the first place! Sometimes the easiest way to get out of the hole is to stop digging.
I think the obvious question is, if you delegate teaching to people who are only limitedly accountable to you, how do you ensure it's happening properly? I vaguely recall that the biggest obstacle is that supervision costs make it fail to save teacher attention on net.
One possibility is to alternate classes taught by a teacher with classes taught by older students (kinda like at university: professor and assistants). The teacher will notice if the kids keep getting something wrong.
You could also alternate the older students, so that if one explains something incorrectly, the other has an opportunity to fix it.
A different possible approach is to have computer exams that ask you whether you have received instruction from another student. If you pass the exam correctly, the other student also gets a small reward. But if many students taught by the same student keep making the same errors, the question is added to older student's spaced repetition schedule.
Strongly agree with this comment as I was thinking of my children’s Montessori experience while reading this review. I would add, though, that Montessori is highly structured it’s just that much of the structure is hidden from the children and done in a way that allows the children to work independently and to choose (within in a carefully designed structure) what to work on and when.
Yeah, maybe this was one of the iterative improvements possible within the well-structured school environment. We just need to teach public schools how to structure the learning in slightly different ways. Encouraging high agency is definitely a goal, not just motivation while in the classroom.
s/was/is/
The essay did say it wasn't optimal, so maybe Montessori just has something to teach public schooling here more than it disproving the general thesis.
Maybe we've got it all backward. Maybe instead of Montessori for grade schoolers, we ought to make high school Montessori-ish, as well as voluntary.
ETA: I'm inevitably thinking of the forced time spend indoors in high school, with so many people.
I'm skeptical if only bc Montessori is significantly more popular for 3-6 year olds than teenagers. If teenagers were more suited for it, I suspect this would have shown up already in the existing market for schools.
It just seemed to me that apart from shop or atuomotive class or home ec, the interior of a school - [and my fairly brutalist in effect if not in design middle and high schools dated from the advent of central air-conditioning in the South so had virtually no windows, and none in many classrooms, which I admit confounds the issue for me] - is not a great way to interact with or learn about the real physical world, nor about one's interests and possible aptitudes within it.
It seems to me that a high school chemistry or physics lab -- even a windowless one -- is a vastly better way to learn about the real physical world than what you seem to be envisioning.
Montessori schools have a high selection effect. They tend to end up with a lot of the no-structure and low-structure learners, because those are the students who succeed in that sort of environment. High-structure learners are less likely to be enrolled in a Montessori school, and less likely to stay once enrolled.
Very possible. Could the "gifted" program at school be cross-grade, and more like Montessori, then? The gifted programs I know of still segregate by grade, and the one for my oldest son is even more isolating (they are given study materials without even changing their classroom)
Not to be argumentative, but did the Montessoris school motivate your children or were they self-motivated?
>Montessori motivated my children
>what children work on are largely self-motivated
My child is self motivated. At public school, he is not motivated. At Montessori he was. This is a sample size of 1 so don't take it super seriously, but..
When I talk to other parents, they have similar great differentiated experiences with their children in Montessori. It's a great structured environment that teaches agency and independence. Highly recommend if you can do it.
I suspect public schools can learn at least something from Montessori schools.
I appreciate this review, but there were a whole lot of places where it was just stated that some particular intervention can't be done or won't work without any evidence. I noticed this most strongly about tracking. Tracking apparently doesn't work or can't be done...until high school I guess where it is much more common (and apparently works/is a good thing?)
This feels to me like someone working backwards from a conclusion and not actually being interested in solving the problems that even they admit exist in todays schools.
It's an interesting reflection to the alpha school review which was incredibly positive, to the point of probably overselling. I see these two reivews as the two sides of the school debate coin.
One side is convinced that large improvements are everywhere and ripe for the taking, and the other side things that improvements are the next thing to impossible.
Many of those points do have links attached. E.g., in regard to tracking, the text is "In the US this is typically called tracking or ability grouping. It’s a complex and controversial topic. The research base is hard to read because there are a lot of ideologically motivated researchers who are either for or against tracking and want to see the evidence a certain way. But the biggest theme in the research is that the effects are small. There are plenty of meta-analyses that find an effect near zero (here’s one example)." and the link is https://journals.sagepub.com/doi/10.3102/00346543221100850
The problem is that the review isn't a literature review, it's pitching a thesis on why school has to be set up the way it is, and is trying to persuade readers of this by telling an epistemically-satisfying story that they can internalize. It does some parts of this successfully, but when it comes time to explain why a tracking-based system doesn't work, it instead retreats to throwing out a citation to a meta-analysis that says tracking in practice doesn't currently shift test scores much. But it doesn't really explain *why* that's the case (our social science isn't yet good enough to answer that with much confidence), and certainly doesn't make convincingly clear why its thesis implies that tracked classrooms can't work, because it's not obvious why they can't provide peer motivation.
So motivation is the key to educational success?
New idea for a startup: instead of training AI to be great at teaching kids, train the AI to be great at motivating kids to learn. Make it create a psychological profile on every student, and do whatever it takes to convince the kid to learn. Some kids will be paid a dollar for every book they read, some kids will be put through stimulating educational games, some kids will be shamed into conformity, some kids will be put against each other to compete for high scores, some kids will be spanked by a robot.
I think the essay strongly implied that AI could never be great at motivating low-structure and high-structure kids. They need the presence of peers doing the same thing.
In other words, "personalized learning" just won't work for most young people.
The essay acknowledged that 1:1 tutoring works, it just didn't dwell on that because you obviously can't do it for every student if the tutors have to be trained humans. Human-level AI could change that.
Maybe I'm assuming too much but it seemed to me that he thought "human-level AI" could never work as well as a "trained human" because, well, it just isn't human. There's something about an actual person that is necessary. I suppose I'm reading between the lines of passages like this:
"One form of learning that has been shown to be particularly effective is deliberate practice. Deliberate practice involves being pushed outside of your comfort zone, focusing on specific, concrete goals to improve performance, and getting consistent feedback. One common characteristic of deliberate practice is that it isn’t particularly fun. Most contexts where deliberate practice is common, like sports and music training, involve expert, individualized coaching. The coach is mostly there for motivation. The coach does other things as well, but the most important thing a coach can do is motivate you to train. Deliberate practice isn’t common in school learning, but it’s a good reminder that motivation is the key to lots of forms of learning in and out of school. Learning isn’t always going to be a ton of fun. In the absence of one-on-one tutoring for every student, conformity is the best tool we have to create the motivation necessary for learning."
I doubt this is because motivation requires learning from someone with an immaterial soul. It seems more likely to be something like, a human tutor actually pays attention to whether you're learning or just tuning them out, and can respond accordingly. Digital learning systems that have been deployed at scale thus far can't do this.
I'm thinking more like "requires someone composed of meat". Seriously.
Though I'd love to see what happens with a digital learning system that does the things you describe a human tutor as doing.
A human tutor can yell at you for tuning out in a way that an AI cannot.
For maximum effect, the tutor should be physically present where the student is.
Does Zoom tutoring not work okay in practice? I thought it was already pretty common.
Correct. If you can't hurt its feelings, it's not a teacher, it's a textbook.
That's pretty much what I was thinking. A human can and eventually will tell you "Don't waste my time", while an AI can't do that convincingly.
I suspect this is also why coaching works to the extent it does. Both that the coach will say "Don't waste my time" and that you will get cut from the team if you do.
I'd also think that a human tutor can better serve as a model of behavior in a way an AI can't. A human child can aspire to be as good as or surpass their human coach, not as easy for their AI one. Operating words here "not as easy," not saying it's night and day or impossible.
We will have to make some radically retrained LLM's if we want to be able to shout at the kids. But I guess that anyway is something we need to teach them before they can become middle managers. Might need to unlearn the spanking bit from the manager bots, though...
Really interesting article. The motivation issue is really what things boil down to when you look at any content online about learning, beit coding, math, music, art, language... the most popular videos on YouTube are all about issues with *sticking with it*, with not *picking up your phone again* instead of doing the things you want to learn.
I don't think learning most things is very difficult if you genuinely put two hours a day of some combination of actual concentrated practice (doing the thing) and study (learning why you still suck at the thing and what you're doing wrong).
There are some smaller issues with plateauing / dunning kruger if you do nothing but practice and zero study-- you end up in the Tim Buckley zone, famous webcomic artist who copy pasted the same lazy face onto all of his characters for like a decade. I think webcomic artists so often plateau in ways that are easily lampooned because they are stuck on a regular schedule to keep their fans happy, and they already have a dripfeed of people telling them with every update that they're already wonderful and don't need to improve.
If you do nothing but study, you have the 'swimming on a tatami mat' issue, you can watch a thousand hours of painting theory videos and not get any better at painting.
(Also, language learning is different and doesn't work how most people think, you need to spend like 80% staring at target-language subtitled telenovelas or anime you can barely understand, or scraping through easy reader books aimed at toddlers, and you have to give these boring activities as close to your full attention as possible for hundreds of hours, so that your brain builds an intuitive understanding of how the language sounds and works in normal situations.)
But as long as you pay mind to those potential pitfalls, **the big big big issue is motivation.** It's *getting* that 2 hours a day of study and practice.
And the only silver bullet I'm aware of, when it comes to motivation, is having a desperate *need* to learn something, a 'sink or swim' type situation where you literally have no other option but to do it (I think this is the environment that Tiger Moms try to cultivate for their kids, and it obviously often backfires when they get older and realize that everything is made up and the points don't matter).
Language learners do this without realizing it as kids when they grow up in a new country, or want to interact with the majority English speaking Internet and just learn by necessity as they spend hundreds and then thousands of hours in that environment.
If you don't have that desperate *need*, then the only other solution is to make sure you have a significant amount of tolerance for frustration and feeling 'low status' because you can't draw a circle or watch an episode of Crayon Shin Chan (think Japanese caillou but funnier and dirtier - yes, it is a kids show) without English subtitles, and make a real serious habit, OR, as the review describes, find yourself a group of like minded peers around your skill level and stick to them like glue, keeping each other accountable.
Otherwise... learning is fucking hard. We've got phones and porn and video games, why would you spend 2 hours a day fucking around in Blender or drawing badly or watching anime you can't understand, making tiny incremental progress towards some imagined goal 10,000 hours in the future that won't feel anything like you imagine it will, once you actually get there?
Interesting that despite coming from almost exactly the opposite perspective as the Alpha school review, they actually have the same conclusion—that secretly the key to school is motivating students. I liked the distinction between high-, low-, and no-structure learners. I’m definitely a low-structure learner, who has always kind of beaten myself up about not being a no-structure learner. Put me in any kind of class and I will do all the assignments and quickly learn the material. But when I try to learn things on my own from pure interest I always peter out after a few weeks at most. The times I’ve had success have usually been when I signed up for an actual class with teachers and other students. After reading this review, I feel like I need to just accept this about myself and sign up for more classes when I want to learn things.
I noticed this too. I wonder at what point "just literally pay kids to do schoolwork" will enter the Overton window of education reform.
Alas, ChatGpt closed that option before we could even start.
I'm guessing that you mean that students could cheat with ChatGPT, and that would remove the incentive to learn the material? If I'm understanding the Alpha school approach properly, I don't think that cheating that way would be easy. ( _Is_ that what you mean? ) If I'm understanding the Alpha approach correctly, (setting aside the home school option), there were very frequent quizzes in the classroom, where cheating would require having access to to the LLM in the middle of a quiz. If the kids are required to put away their phones during the school day, and not fire up an LLM tab in the middle of a lesson+test, this should be feasible to prevent.
Very much agreed!
>I wonder at what point "just literally pay kids to do schoolwork" will enter the Overton window of education reform.
is exactly counter to
>But we will do our best to keep them moving along with their peers because _that’s the best idea we have_ to motivate them.
[emphasis added]
The Alpha school results strongly show that, yeah, we _do_ have a better option for motivating learning.
( In retrospect, having read both reviews, it is a pity that the two authors didn't write a combined essay on both lines of evidence. )
Same! I'm really curious what the publisher of this piece would think of that facet of Alpha School!
I've been wondering myself, also, if there is some way I can apply the lessons of motivation to myself in doing online schooling.
> they actually have the same conclusion—that secretly the key to school is motivating students
The difference is that the Alpha school review concludes that kids should be motivated by learning opportunities and rewards, while this review concludes that we need to sacrifice the learning opportunities of the no-structure learners to motivate the high-structure learners by their presence. The effect on high-structure students may be comparable, but the effects on no-structure students are dramatically different.
Yup! And, bluntly, the no-structure students are _important_ (making the assumption that AI doesn't totally take over). They aren't _all_ of the future decision-makers, nor are _all_ future decision-makers no-structure students, but I expect there to be a lot of overlap, and, for our society as a whole, it help a lot if future decision-makers are well-informed.
Great review! This crystallized my own observations about school and my place in the rationalist community. The rationalist community is filled with no-structure learners like Zvi Mowshowitz and Eliezer Yudkowsky, and consequently much of the prevailing wisdom doesn't work as well for low-structure learners like me. For instance, I spent a while trying to read textbooks on my own like LessWrong perennially suggests, and I made some gains but lost interest once the material got uncomfortably difficult.
One thing this review doesn't go into that I definitely observed in my own education is that accelerated learning can be pretty traumatic for someone who's used to not needing much structure and is suddenly thrust into a situation where they need to avail themselves of all the structure just to keep up.
The review is great at describing the needs of the learners who needs structure. But the suggestion that the no-structure learners should be sacrificed to them is triggering as fuck.
I need to calm down, and maybe in a week I will be able to write a reply that will consist of something other than screaming...
Really happy to see this one made it.
I believe Scott has written at least two posts about school that I've read. Between those and the discussion it spawned it my impression was always that many around here (and on reddit) experienced the downsides of that rigid structure and none of the upsides. They then concluded that obviously compulsory schooling must be bad and evil, no qualifiers.
My own experience was markedly different. I have, at many points in my life, required external motivation for something I knew was good for me but that I couldn't motivate myself do. Much more so when I was younger still. I'm extremely convinced school would've been in the category "I should do that, it would be nice, but, oh videogames!". Looking back I'm very happy I was made to go.
A lot of the discussion around here has been around largely self-motivated people unhappy with how much school got in their way. I'm happy this review directly addresses that this is true, it does exist, but it's not malice, it's a tradeoff. And then goes on to explain that in detail.
Already waiting so read all the rebuttals and why, actually, school is just a socially acceptable way to imprison small humans who don't have the power to stand up for themselves.
Education is one of those things that everyone has an opinion on, because we were all educated in some way, and for the vast majority of people posting here, it was probably in schools. And as you point out, the typical posters here are not at all representative of the median American.
The simple fact is that compulsory education isn't going anywhere anytime soon, because parents need childcare. Unless an AI robot tutor/nanny can stay home with a five-year-old, it makes sense to have schools for kids to go to while their parents work.
Nannies would be cheaper than teachers.
Yeah, but would a nanny have 30 kids to supervise?
Nannies aren't required to know anything needing teaching, just to supervise children. This will always make them more efficient when the goal is just to supervise rather than teach.
I think the "meeting student where they are is a failure" part needs its own article/review, because it's repeatedly relied on by this article but I don't think it has a strong foundation at all. Also it doesn't seem to pass my priors. From my own experiences, it's obvious that grouping students that have the same characteristics would be better than having too wide differences. It makes me starts to skim this review to get to the point fast, unlike previous review where 18k feels short. I guess the conclusion is generic enough that it shouldnt be 100% wrong, but I don't think the arguments really support it or stamp out denying questions.
I liked that the author put the thesis near the beginning. I'm not quite sure why—I think it helped me read better somehow. I wish more people would do that.
It’s a basic principle of good writing - put the thesis near the beginning, so that you can read everything in the context of where it is supposed to lead.
And yet not all reviews in previous years followed this principle. Some of those reviews even won!
Scott gets a lot of mileage out of not following this principle. In many of his best articles the reader is being taken on a journey with no inkling of the destination.
Which ones do you have in mind? That might shed light on when this technique is and isn't appropriate.
One tangential point is "The common school movement emerged between 1830 and 1860." That's about the time of the Jacksonian Revolution, when the political philosophy of the United States went from a "republic" where the voters were the highest social classes and expected to wisely choose for everybody to a "democracy" when the voters were (conceptually) all adults. (Of course things weren't so pretty in practice, but there was a substantial broadening of voting, moving away from the idea that you had to *qualify* in some way.) There was also a high level of immigration. So getting everybody to think in generally the same ways about *everybody* as the body politic was starting to become important.
I like this review a lot. It does a good job arguing against the Alpha reviewer’s (even at the time obviously) crazy claim that the new system could work for 30-70% of students. But it does very little to persuade me against the weak-Alpha argument, which I read as mostly “hey, look at this awesome new system that takes the top 5% and helps them learn twice as fast.”
Honestly, I don’t think the author of this review would have any problem with that claim! For no-structure and maybe even some low-structure students, it’s a really excellent model. But this is definitely a good reminder of the commonsense truth that many people just suck at learning, and old-fashioned schools are set up to teach even them to read.
All that said, I think this author is too gloomy (or maybe just not excited enough?) about increased tracking. It’s true that many high schools already give talented students some opportunity to jump ahead, but it really tends not to be much, and there’s certainly a prevailing attitude where parents and administrators try to push anyone into advanced classes they can.
I’m not sure my experience generalizes super well, but I can tell you there’s at least one school where:
- Everyone takes the International Baccalaureate curriculum.
- Entry to the school is decided by lottery.
- Some students are encouraged to drop out when it becomes clear they can’t handle it.
- Many (*many*) others are not.
- Classes move at a snail pace and no-/low-structure learners (*cough* me and most of my social group *cough* maybe 10-20% of the school *cough*) get bored and play Geoguessr in Physics.
- The teacher doesn’t care because the smarties are still scoring at the top of the class on all the very-slow-moving quizzes.
- IB exams roll around and the school performs *terribly.* I mean, seriously, one of the worst in the world. We have *never* had a *single* student receive a 7 in Global Politics.
- Even the best students tend to be unprepared. We haven’t got results back yet, but I’m pretty sure I shat the bed on Chemistry and maybe Physics, and I have (genuinely smart and talented, way beyond my own ability, etc.) friends who shat on Math, Physics, Spanish, Biology, …
This is terrible! Obviously there should be more tracking and separating done, and it’s true that at many high schools there is, but there also are *plenty* where this is the story. AP/IB classes full of confused, slow, distracted, and distracting learners who mess things up for everyone else.
And even if most high schools aren’t like that, nearly every middle school is. I’ll admit it matters less in elementary, where everyone’s fine just playing with blocks or whatever, but most smart young people I talk to have middle-school horror stories. Bullying sometimes, but mostly just rowdiness, unpleasantness, loneliness, and boredom. School as an institution is probably good on measure—but that doesn’t mean we can’t make it massively better for the kinds of kids who’d be into Alpha.
Even just based on my own experiences as a student I think I agree with a large fraction of the claims made here. But, there doesn't seem to be any attempt to account for the costs of the incumbent approach, or at least there seems to be an underlying assumption that the costs are worth paying.
One is the cost of holding back the students who could be learning more. What would actually happen, in aggregate, if we let some fraction of students fail to learn anything, in exchange for letting some other fraction of students learn several times faster? What would happen if we let students choose to ignore (and fail to learn) subject A, but excel far beyond 'grade level' at subject B, until such time as they find a reason to put in the effort in subject A? I have no doubt this would cause some problems, but it would also do things like create a highly motivated population of scientists and engineers and doctors and lawyers and artisans whose careers are 10-20% longer than they would be under the current model, and they'd enter their most productive years with an extra 5-10 years experience under their belts.
Another is the cost of causing mental and physical health problems. It's a cost to force kids to sit still for hours in a row, and restrict their ability to go use the bathroom or move their bodies, and to crush their desire and motivation to do various things and explore various interests that would, for them, be valuable. It's a cost to force kids to be in the same room as people they can't stand or who treat them badly, for months or years. It's not clear how much, now or in the past, the loss of agency school demands is worth the extra learning some proportion of students will get in return.
And of course, there is the direct financial cost. K-12 schools employ 8 million people in the US. How many fewer teachers and staff would we need if most schooling were replaced with other forms of instruction, AI or MOOC-like otherwise? At the cost of how much learning, for how many students?
Relatedly, what would be the learning *benefit* of such replacement be, if it enabled people of any age to choose to start learning any subject at any time, without fear of social sanction for seeming to be 'behind' or in 'remedial' classes, or needing to go a bit slower, or wanting to study something unusual? What is the cost of us having this societal idea that a student is something you *stop being* after a certain age, unless you go 'back to school,' at a fairly high personal cost - of seeing education as a separate magisterium from the rest of life?
Edit to add:
I've spent the last few years RVing full time. I don't have kids, but I've met (and read/watched content from) a lot of families that roadschool their kids. What you can't avoid noticing is that 1) They tend to sign each kid up for different forms of learning depending on their personality, because *that is what best motivates each kid to learn.* There's a lot of options out there these days. and 2) The kids themselves seem to be more mature, independent, and worldly on average. But also 3) Sometimes the kids express a desire to go (back) to regular school, and when they do the parents should, and often do, listen.
I agree strongly with this comment, and I disagree strongly with the fairly common view that because high performing students will do well in school and life anyway, it's okay to waste 70% of their time in school and several years of their life on a system that isn't designed for them.
To a small extent, the UK is closer to a personalized system (and most of the commonwealth countries have the same system). Students are placed into specialized tracks of their choosing at age 13. And people complete their law degrees, medical degrees and PhDs on average 2-3 years earlier.
It has it's costs though. It's very difficult to switch tracks and students are less well rounded. I knew some brilliant science students who didn't know what apartheid or communism were.
The U.S. public schools spend roughly 3x more per pupil today than they did in the 1950's, with no clear evidence of better results. So even if we don't tear down the fence, maybe we don't need to maintain such an expensive fence.
I know people who, thanks to tracking and gifted education, were able to graduate high school a year early or finish college in three years thanks to AP credits, or both. Isn't this a good thing.
I was disappointed that the effect of parents was never discussed. I still remember working with my mom regularly through high school; she provided detailed comments and corrections on every paper I wrote. I did the same for my daughter and included math / sciences in the interactions. We both became very successful technologists. Friends tell me that the parents who attend parent teacher conferences typically have successful kids. Perhaps we need to improve the education results by making the parents more responsible for success?
How would you propose to do that?
Kids don’t get to play football unless the parents do certain things? Require volunteer parent activities? Recognition for outstanding involved parents?
Some of it is just expectations. I think the emphasis is on “schools must educate the kids” rather than “parents must educate the kids and the schools will help”.
The parents you need to motivate are the ones who aren't already very invested in their kids' academic success. None of those ideas sound like they would work on those parents.
If you think it’s hard to motivate students, who actually have to be in the place where you’re working, it’s going to be even harder to motivate the parents, who spend almost no time there.
One additional thought on why MOOCs have such a low completion rate: I started listening to one about ten years ago, and the first ten minutes was the professor covering minutia about attendance, grading, etc. It was literally just a filmed lecture with no edits. I very quickly returned to buying college lectures from The Teaching Company.
There exist MOOCs that are better-designed than this and their completion rates still aren't all that high.
What if the top 5% were taught to maximize learning and the rest to maximize motivation? The layman can benefit a lot from civility and motivation. But one Einstein can change civilization.
The education system in America is socialist and oddly enough public schools at communist countries seem better at identifying and supporting outstanding academic talent at an early age.
Such a system for public schools was tried in Britain in the 1950's. Students scoring high on tests was packed into special schools around age 11-12. I think there was a huge back clash when upper class parents realized that not all their children were making the cut. Only when the elite schools are expensive do we accept them...
Very interesting review. I do want to take partial issue with this:
>There’s ample evidence that we can’t actually teach critical thinking divorced from content.
My issue with that statement is that people often incorrectly infer therefrom that students must master or memorize content before being asked to think critically. But that is not actually true; a student can "analyze Han and Roman attitudes toward technology" using the set of documents here: https://secure-media.collegeboard.org/apc/ap07_world_hist_frq.pdf -- including using evidence to support a thesis,; analyzing how the evidence might be incomplete or misleading; and what additional evidence might be useful to test the student's thesis.
Of course, a student with lots of background knowledge will do a better job than a student without background knowledge, but that background knowledge is not necessary for the student to become more proficient at higher-level thinking ("higher level" being a reference to Bloom's Taxonomy https://citt.ufl.edu/resources/the-learning-process/designing-the-learning-experience/blooms-taxonomy/)
Yeah, it shouldn’t be learn some content and *then* learn to critically think about it. Rather, it’s that you learn to think critically as a byproduct of learning some content. You have to be aiming at the content in order for the critical thinking to come.
It’s like how you clear your mind in yoga as a byproduct of paying attention to your breath, not by first paying attention to your breath and then clearing your mind.
>Rather, it’s that you learn to think critically as a byproduct of learning some content.
I don't think that is quite right. Critical thinking does not automatically come as a byproduct of learning content; it must be independently learned. It comes from addressing questions above the level of mere recall and comprehension. Eg https://facultycenter.ischool.syr.edu/wp-content/uploads/2015/10/Revised-Blooms-Questions-Starter.pdf
Right, I should be more careful. It’s not a mere byproduct of learning content - it’s a byproduct of learning content in particular ways. But there has to be some content that you’re learning.
This review believes, but doesn't quite state explicitly: the purpose of school is to maximize motivation *for the high-structure learners*. This review seems to be telling us that we have to put everyone in the same classroom because this is the only way to get those high-structure learners to learn. I think the author would agree with a statement like: "we're going to sacrifice the learning of the top X% of the student body, because teaching them really slowly in the same classroom is the only way to keep the bottom kids learning at all."
The review at one point discusses something called "leveled reading", but dismisses it, because it "reduces the achievement of the readers who struggle the most". Great, but what does it do for the other readers? We never find out. (There's a link, but the linked article has no numbers in it.)
I can sort of understand this as a philosophy, but also I can completely understand why anyone with a smart kid would try to pull their kid out of it.
I’m pro-smart kid too.
Because “high-structure learners” suck at learning, all schools must be made dumb and boring for everyone. No Child Left Behind because we are racing to the bottom.
For me school was demotivating so I'm not buying any of this. I learned a lot in this life but none of the things that stuck with me were learned in a classroom.
I enjoyed the essay, I agree with other commenters that the timing of this post works to its benefit.
I've noticed a number of comments so far suggesting different interventions to improve motivation. My read of the education literature (not an expert but did take a PhD field course in Economics of Education) is that basically almost every intervention under the sun has been tried and all of the ones that showed promise have disappointed once the programs have been scaled up. This includes interventions like paying students, paying parents, giving parents more information, behavioural nudges, etc.
I think one thing that tends to go underappreciated by those of us no longer in school is that genuine "learning" actually is fairly uncomfortable/hard. It's difficult to consistently stay outside of your comfort zone and encounter failure. Just think of how little "new" material the average adult actually learns once they're finished with their schooling.
On top of this I believe most adults have an inflated view of their own ability to learn. They equate learning some new procedure/process at their workplace to learning new content in school, but the two really aren't comparable. This has been made worse by graduate programs that are aimed at working professionals but still award a degree (i.e. MPP/MBA/etc.). These programs are often cash cows, they involve very little learning for most of the students. They tend to contain courses with 2nd year undergraduate course material, but the grading is done on the inflated graduate course curve. As a result, graduates end up getting a false view of what learning is really like.
This was a wonderful review, and as many commenters have said, a nice contrast to the Alpha School review.
A point I was waiting for that never came, was that the same student can be no structure, low structure or high structure in different domains. And that students can even move between categories over time.
Speaking for myself, I was solidly no structure at mathematics (I participated in the international math olympiad almost completely with self study), and high structure at writing - I just had no motivation. That was, until I took a philosophy of religion course in college and was able to write about things I actually cared about, and had opinions on. Suddenly I was no structure and able to ace writing assignments while actually enjoying writing.
And so, I'm unpersuaded by the assertion that a more personalized approach can't work for a typical low to high structure student. We haven't found an approach that consistently works, but that doesn't mean it doesn't exist.
This is close to my experience. I was good and self-taught at all the topics I was interested in: maths & sciences, really. I was bored in languages and literature… until I found some I was interested in (mostly philosophy and classical studies) and it turned out I was good at writing too.
This is what I find about getting philosophy grad students to learn math. A lot of them never thought they were good at math, but once they see that a philosophical question that they care about depends on something mathematical, they start to pay attention to that mathematical thing and start to improve.
>once they see that a philosophical question that they care about depends on something mathematical
I'm curious, what sort's of mathematical things do they find philosophical questions depend on? My knee-jerk guess is Gödel's incompleteness theorems and maybe Cantor's diagonalization argument??? Other examples?
My work is in formal epistemology, so I'm often showing that thinking about concepts like knowledge, and justification of belief, works better when you think about it in terms of probability. I personally was most interested in algebra and logic as an undergrad, and not interested in real analysis, but ended up teaching myself a lot more real analysis when I realized how central it was to thinking about conceptual issues for probability.
But you get other things in other areas of philosophy.
All of the classic diagonalization arguments (I count Cantor, Tarski, Gödel, and Turing as giving versions of it) end up being relevant for lots of thinking about meaning, mind, and language, as well as for understanding mathematics and computation. And in my department, a lot of people learn to code because they realize that making computational models of information flow in scientific communities is going to help them address the questions they're interested in for scientific epistemology.
Many Thanks!
Seems to me the real question is 'why did we spend so much time solving quadratic equations in school?' Really -- who cares what the solution to the equation of a parabola is? Are people continually coming up to you and saying, quick, quick, where is this parabola equation going to cross the y axis? Yeah, right, never. We spent so much time on it in school was because it was complicated enough that you had to spend some time learning it -- ie, we learnt it because it was a great way to use up time.
I’m sure you've seen the SMBC comic.
Student: Will we ever use any of this algebra?
Teacher: You won’t, but one of the smart kids might.
A thoughtful antidote to reform enthusiasts. Still I'm a bit puzzled. On the one hand schools exist to maximize motivation. But it seems that none of the three groups you've identified have their motivation enhanced. The unstructured are bored, the low structured develop a few habits but never catch fire; the high structured group loses steam the longer they stay in school. Many grow to hate school and learning. So whose motivation is augmented?
If I am told, "Do this unpleasant thing or I will cause you pain", I have not had my intrinsic motivation increased. But I have had my motivation to actually do it increase. He's talking about the second meaning of motivation.
I was surprised by the claim that deliberate practice doesn't work. Maybe we just aren't doing it right.
Sports are incredibly popular as a leisure activity. And during sports, even at an amateur level, you are generally being pushed to the edge of your abilities against people of a similar skill. You're constantly adapting, learning and then trying again with improved skill and knowledge. That's deliberate practice.
Several other popular leisure activities have a similar structure - chess and other board games, online pvp video games. Even solo activities like climbing or art where you are challenging yourself to do better rather than competing against another person.
Getting into a flow state can be very motivating. But it's tricky, it only works when you have incremental goals that are just within reach. If you push too far beyond that, then it becoming demotivating.
If I'm understanding the reviewer aright, I don't think they're saying that deliberate practice doesn't work — just that it's really hard to do it, b/c deliberate practice is hard and painful. (For what it's worth, K. A. Ericsson — the progenitor of "deliberate practice" as a concept — wrote in the book "Peak" that there's an only-slightly-inferior move he dubbed "purposeful practice" that is much easier to do in schools. Less autonomy needed.)
One minor correction - the text says that the high-structure students are the ones who are constantly in office hours. This is definitely wrong - the high-structure students are the ones who *should* be in office hours, but the fundamental problem is that there’s nothing that makes them come.
I have far more formal education than most barn cats, but a typical disdain for academic credentials. Beethoven never got an MFA. The only use I can discern for art school is so British musicians can drop out of it.
Still, how many of us insist on a self-taught brain surgeon? A Khan Academy trained aerospace engineer to design our airplanes? A really motivated amateur to operate a power grid?
Smh as a motivated amateur operating a power grid.
Or that dude who did a couple of lobotomies on stray dogs for practice now removing your brain tumor.
Great article. I see that many commenters have said that giving more opportunities to high-performing students is not necessarily a problem. I enthusiastically agree.
In England, in 1944, the government set up a tripartite system where 11-year-olds took an IQ test, called the ‘11+’. The highest scoring 10%(ish) were sent to a grammar school which aimed to send those kids to university. The rest went to a Secondary Modern, which taught less academic subjects.
By the 1960s, it became obvious that this system was too successful, and society was becoming divided into middle-class kids who passed and working-class kids who didn’t. One unexpected problem was that the smarter working-class kids were passing, which left fewer smarter working-class kids to represent them in politics etc. The government shutdown this ‘selective’ system in the 1960s and replaced it with a ‘comprehensive” system.
Two counties stuck with the selective system and I lived in one. I passed the 11+ and went to a grammar school. I’ve had a very successful life. My brother failed the 11+, and went to work on a building site. He has had a very successful life too. It would have done him no good to stay in school for more years, learning calculus and Shakespeare.
I wrote more about this here:
https://raggedclown.substack.com/p/11
I reviewed this before it was promoted, rating it low.
I have a variety of complaints, but the main issue I had was, I felt that the concept of "motivation" as pushed in this review wasn't actually well defined, and that this was used to push a class of Motte/Bailey. It starts by stating that schools aren't really about learning, and presents "motivation" as the new optimization target. It quickly asks "Motivation for What?" but the section doesn't actually address this question at all. It's roughly implied that it actually means "motivation to learn", which in turn maximizes learning.
Which ruins the entire thesis, because now he's back to arguing "best of all possible worlds" and "actually, this is the best way to accomplish learning". But, he doesn't present evidence for this case and actually, the evidence is starkly opposed to this case. You can't just handwave all of the evidence saying that traditional American schools are not very effective, and then conclude that traditional American schooling is the best possible education system.
Which is all to say, I don't think the author is actually being coherent, with the primary culprit being, replacement of standard terminology with non-standard terminology, then switching between the meaning of the standard terminology and non-standard terminology as necessary to win the argument.
"Age-graded" is mentioned a lot in this review. In the past, older students taught younger students in a "one room schoolhouse", with adult supervision. Maybe it doesn't scale, or doesn't develop motivation as well?
Oldest siblings have stronger personalities, are more of leaders (my observation). They practice being leaders with their younger siblings. They could (hopefully often do) care for their younger siblings as they lead them. They are sometimes a bit, or more than a bit, pathological. Optimistically, they just don't know better. We don't teach leadership and being responsible for other people in school (proportional to learning "material"). Youngest siblings don't practice leadership and care with younger siblings. Maybe it would be good to teach leadership and care, since many or most people end up being leaders to some extent, formally or informally. Having older students teach younger ones could help with that, perhaps. Also, perhaps teachers can delegate some of their workload to students. Students and teachers then would have more time for one-on-one tutoring for those who need it.
Great piece. While you allude briefly to this, I wanted to emphasize it more: students aren’t all one thing. I was a no-structure learner when it came to reading and writing. I could have learned those skills alone on a desert island. But I was a low-structure learner when it came to math. I needed somebody to walk me through it. So I needed school. I think a lot of students are like me; they might not need formal schooling for one area/topic, but they do for others. Some students need it for everything, but I think it is the rare, rare person who is no-structure across the board.
There are different types of "structure":
- when you keep nagging and threatening someone to study, because that is the only way to get information inside their head (and the person keeps fighting you all the time),
- when you help someone overcome distraction and procrastination, by giving them a time and place to learn (and the person may be annoyed for the first 5 minutes, but then enjoys it),
- when you give someone the right books and tell them the right words (and the person enjoys the help from the start),
We should distinguish between kids who actively resist learning, and kids who want to learn but need some kind of support. Both are "a structure", but of a different kind.
A pretty good review which rhymes very well with my own experience on the other side of the Atlantic.
I am starting to come up with a theory that normal schools actually are there to impart 'common knowledge', the knowledge that we know others know. Kind of like what you gain from an oracle saying at least one person have blue eyes.
The schools as we know was originally made so nations could organize their economies for war, which required some kind of common outlook on things.The countries that didn't got colonized by ones who did, and they then copied their own school systems.
I wonder what a world where we to loose this kind of common knowledge would look like. Right now everyone understands 'in democracy you have to hold fair elections', most people have heard of separation of powers even if they can't quite explain the concept, and some few people know about comparative advantage in production. Let us imagine this moved one step to the right, so most people had heard something about elections and only 10% knew courts are supposed to be independent of the normal government. Even if the smartest people were even smarter than they are now, I still think society in that world would be inherently unstable.
I think that this piece gets a lot of things very right about schooling. It's refreshingly clear and close to my own view. (Though it does in fact seem to be making the case that school is maximizing learning, when you take low- and high-structure children into account.)
For further evidence that age-graded schooling is doing something important in modern society, I'd also point to the lack of interesting variation among countries in terms of how they educate their kids. You don't really get the sense that there are competing systems out there.
I do think that there's a danger in slipping from "here is why things are the way they are" to "things HAVE to be the way they are." The piece makes some caveats in this direction, which I appreciated.
That said, I do find myself thinking that if AI tutors ever get as good as some people think they can be, we might figure out interesting ways to make it feel like "we're all doing the same thing" in the classroom while students are getting personalized tutoring. My students often love doing a bit of lightly-personalized practice with software (Deltamath). It's not that hard to imagine a better version of that software that teaches extension content during that block of time.
And at the college level, would it be that weird to have a "lecture" where the professor is setting the stage for a personalized lecture? Maybe you are assigned to a large biology lecture, no matter what bio class you're taking, so that the students present are taking many different courses. Class begins with a brief talk or exercise on a topic of broad interest, and then out come headphones and computers for some magic robot tutoring.
This is all sci-fi at the moment, and honestly I think it will be for the forseeable future, but I do wonder if there are other motivational structures that are possible.
>> "It's refreshingly clear and close to my own view."
Agreed! Michael, could you share how Deltamath "lightly personalizes" math practice?
>> "And at the college level, would it be that weird to have a "lecture" where the professor is setting the stage for a personalized lecture? ... Class begins with a brief talk or exercise on a topic of broad interest, and then out come headphones and computers for some magic robot tutoring."
I wonder if an even more motivating (just to stick to the theme o' the review) version of that might be that the professor opens up with a simple (but big, conceptually rich) question, then students go to their AIs for a fast, individualized dialogue to try to figure out the answer. (Something like this is what I do with my science classes for gifted kids, except minus the AI.)
I agree with the reviewer that if you define “personalized education” as completely self-directed learning then it will certainly be unsuccessful as a K-12 core academics ed model for almost every student. You cannot just stick some IXL loaded iPads in a room and hope the kids decide to work on math. But I see no reason why customized instruction, practice problems and testing for a specific student’s level must mean a self-directed learning environment with no rewards, accountability, structure, public visibility or recognition. This doesn’t have to look like someone signing up to take a MOOC during their spare time on a whim.
At one extreme, personalized ed may not be self-directed at all. It could be an academic tutor assigning you daily learning modules to work on, setting deadlines, handing out grades based on your performance relative to their goals, etc. The difference from regular school being that your assignments are customized to you - you have a tutor not a mass lecturer. The promise of today is that technology can enable a tutor working together with well-designed apps to coach a substantially higher number of students at the same time than would be possible in paper-and-pen days.
My impression is that Alpha is somewhere in the middle, allowing the students (perhaps with parents?) to set some learning speed goals and make some choices. But clearly there is a lot of accountability with all the testing. And they have a rewards systems they are developing with some controversy. It is interesting that the Alpha review independently came to the conclusion that their motivation system was a big part of what is driving success with that school model so far. I am sure much can be learned from sports, music and other non-academic activities about motivation models that don’t involve putting a bunch of kids of mixed ability in a room together to learn in lock-step. The Scouts use things like badges, for example.
One thing I didn’t understand from this review is how all the students being the same age was part of the motivation of the current school system. I understand the practical logistics behind this design, but how it relates to motivation was not clear. Is it more motivating for an 8th grader to sit completely lost in a math class watching same age peers easily answer the teacher’s questions than it is for the same 8th grader to sit beside a 6th grader who is equally unsure how to find an LCM?
This framing of school reminded me of Mike Saint-Antoine's recent article on military training:
https://www.mikesblog.net/p/how-to-make-a-tribe
Awesome insights!
My current beliefs after reading both recent entries on education:
1) I still believe that we could get much better at educating our top quintile of students.
2) I think economic growth and breakthrough innovation comes from the top tier of this top tier
3) I think the 80% could learn just as well or better and for less if we focused on getting them the basics and allowing them to shift to an apprenticeship type model where desired.
4) I think AI coaches supplemented with group activities and human teachers will lead to amazing breakthroughs in the near future (for self motivated learners)
5) I am extremely skeptical about the ability of the current education establishment to redesign how education can or should work. They and their way of thinking are more of a hindrance, on net, than a help.
I am probably wrong on one or more of these though.
I went to a selective high school that had internal tracking in every mandatory subject: math, science, English, and social studies all had advanced and regular classes. Students who were in one advanced class were typically in advanced classes for all those subjects, including me. One year I was dating someone in a regular English class and attended it with him during a free period (as one does). I was surprised to see that there was daily homework, about a paragraph of writing per night, plus regular quizzes on the content of the readings. The class didn't seem to be doing well in either regard. I was also a bit surprised by the silence of the class, and how happy (or at least accepting) the teacher seemed to be that I was participating in the class despite not actually being enrolled.
In advanced English classes we wrote 1-2 essays per book (with no homework most days besides reading and having thoughts on the reading) and the conversation never stopped and infrequently went off-topic. Occasionally we finished books ahead of time because everyone had read ahead. Similarly, in advanced science classes, students often read unassigned material and asked teachers for additional conceptual guidance, sometimes after class. (Teachers also seemed to be nicer, and more interested in helping interested students, in the advanced track. Regular track teachers were more likely to have reputations for being mean.) None of this stuff was graded, whereas each daily bit of homework in regular classes seemed to be an effort to tip students over the passing line.
It's clear in that the advanced track served students who needed less structure at least in those subjects (if not overall), but it also seemed that the regular track served to whitewash the performance of lower-performing students. "Look at these 18 paragraphs they've written this semester", you could say, "they've done 65% of the work." And maybe that adds up to 65% understanding of the books read or narrative devices discussed or self-expression opportunities afforded.
As a 'low-structure learner', I've found that one of the reasons motivation is hard without structure is that plenty of entities are willing to make you spend a lot of time on bullshit – if they aren't paying anything for a teacher. I've taken stupid classes, but I've done a lot more stupid online modules. Teachers are expensive, and that's a honest signal that someone has at least some real investment in teaching you something worth learning.
The most promising alternative to public school that I've seen is a thing called modular learning -- at least the version developed and practiced by Jeremy Howard (tech genius -- https://en.wikipedia.org/wiki/Jeremy_Howard_(entrepreneur)) and a woman named Manisha Snoyer. It's homeschooling done in a way with some structure, curated materials, and some shared online classes and learning experiences with other families who are taking the approach.
Website: https://www.modulo.app/about-modulo
Substack: https://teachyourkids.substack.com
I'm not suggesting it as a solution for the problem of how to educate all kids -- though who knows, it might work -- but it does seem to me like a flexible approach that could be adapted to work for many kids.
Here's a podcast interview with Jeremy Howard where he talks about how he and his wife ended up homeschooling, and his observations of how his daughter was affected by different learning environments, and his own developing theories of how kids learn and what helps and what hinders. https://podcasts.apple.com/us/podcast/jeremy-howards-journey-from-traditional-schools-to/id1702759553?i=1000629636175
School isn't quite like Chesterton's fence where it's been around so long we forgot why it was created. Instead (as you note paragraphs after invoking the fence), mass education is relatively recent. The reason people are often impressed by old timey written tests is because a smaller fraction of the population was taking such tests.
IQ is increasingly heritable with age. Just as school effects fade out during summer vacation, so do such effects when students eventually graduate from school. Thus gains from school are temporary. Per James Flynn, gains in IQ over time don't appear to be gains in g either, so it can be difficult to generalize effects of such gains from other effects of IQ.
> Classic studies like Asch’s line experiments, where 75% of participants denied obvious truths to match group answers
75% did that at least once, but that's out of 12 incorrect answers from a larger set of 18 and "The average participant gave in to the group three times out of 12" https://www.overcomingbias.com/p/conformity-mythhtml
> Schools spend a huge fraction of their resources on special education, providing the structure and systems that those students need.
Why? They aren't a huge fraction of the students, and if we think of spending as an investment, there isn't that much of a return for it. The focus should be on cutting costs since there is such little benefit. Similarly, the most disruptive students might be abandoned for the benefit of the rest https://www.overcomingbias.com/p/helpful-inequalityhtml
> Maybe the best way education can contribute to all of these causes is to focus on providing a broad, basic education to as many children as possible, and maybe our current system is the best idea we’ve had so far for doing so.
What is the reason for believing either of those things? As you acknowledge, the later purposes were added after the system was first created for the earliest purpose, and it would be astonishingly lucky if it were the case that it just happened to serve those later purposes it wasn't originally designed for. Recall also that the original political purpose of shoring up democracy preceded the rise of schools of education and critical theory such that my mom was required to study (distillations of) Pedagogy of the Oppressed to teach Special Ed. There isn't just conservation in the school system, there is also rot https://www.overcomingbias.com/p/what-makes-stuff-rothtml
My problem with this review, which could well be right in some respects, is that there is a lot of speculation without appropriate empirical evidence. For starters, much of the evidence cited on the effects of tracking does not even measure tracking in the sense that many ACX readers will be thinking of and the few I know of that do tell a slightly different story.
If you open the linked meta-analysis and look at some of the individual studies, you will see that the studies on overall achievement typically look at the effects of separating students at the high school level into different schools where they often learn totally different curriculums, relative to comprehensive schools. However, those comprehensive schools track their students according to ability internally. For example, every UK comprehensive high school (11-18 for foreign readers) continues to group their students by ability, but a study like Hanushek and Woessman (2006) is treating us as stopping tracking when we move from grammar schools to comprehensive. All that really happened is that we abolished the substandard secondary modern vocational schools. I never shared a Maths, English, or Science classroom with anyone in the bottom two thirds of the ability distribution during high school, but all these studies will say that I wasn't tracked. Theoretically, the move to comprehensives could improve the ability matches between students and their classroom peers. If some students are late developers or richer parents game the admissions system, you could have bottom half ability students stuck in the higher school who get left behind and top half ability students stuck in the lower school with no opportunity to progress.
Meanwhile, the randomised trials that I am aware of that actually measure tailoring the classroom according to ability show very different results, with the caveat that they are all from developing nations. One set of evidence comes from the teaching at the right level randomised trials in India (link: https://www.nber.org/papers/w22746). The cleanest example here comes from Haryana where they grouped students together by ability across year groups for an hour a day for a whole year for Hindi lessons and students progressed by 0.15 standard deviations relative to control. That is likely close to a whole year's typical learning (source: https://blogs.worldbank.org/en/impactevaluations/simpler-way-communicate-learning-gains). The Uttar Pradesh intervention did make use of outside volunteers in their teaching alongside ability grouping, but the effects were four times as large! There have also been computer-assisted interventions for Maths that target students by ability level to substantial effect (https://academic.oup.com/qje/article/122/3/1235/1879525). The other example comes from Kenya where Duflo, Dupas, and Kremer randomly assigned schools to track students by ability for 18 months and found that learning improved by around 0.2 standard deviations (link: https://www.aeaweb.org/articles?id=10.1257/aer.101.5.1739) - about a year's extra learning again. This study, although likely not subject to publication bias, is a bit weaker as it's merely p<0.05 rather than p<<0.01 like the Indian interventions.
I'd also add that I think there are other areas that I think are poorly evidenced or not evidenced. The three things that stand out are the claims that education quality doesn't matter much, learning loss wasn't that heavily driven by length of school closures, and that you can't group by ability rather than age group. The first isn't true between this study of the UK that also outlines 4 US studies in Table 1 (https://ifs.org.uk/sites/default/files/2023-09/WP202324-Exploiting-discontinuities-in-secondary-school-attendance-to-evaluate-value-added_0.pdf), there are meaningful causal differences in quality between schools. Additionally, the school vouchers study mentioned in the deBoer package is about sending poor students to wealthy private school going badly, which could very well be due to them being low ability students there who can't keep up with their peers. I quibble the learning loss point because the length of school closures was enough to fully explain learning losses in the one study I have seen (https://www.aeaweb.org/articles?id=10.1257/aeri.20210748). My uncertainty on the unevidenced age group part is that i) grouping by ability rather than age did work in Haryana and ii) you can also spin arguments about why doing so may improve student effort. Maybe students falling behind would work harder to keep up with their friends who are moving up, or low ability students would stop acting out in an effort to show off if the only people they can show off to are several years younger. I don't know because it's almost never tested.
Finally, I'd also throw significant doubt on the likelihood of the linked meta study to even measure the thing it is measuring accurately when it cites Manning and Pischke (2006) as evidence of the effects of tracking (in the separate schools, not classes sense) on education inequalities when just reading the abstract of their paper would tell you that their argument is that you cannot reliably estimate any causal effects using the available data.
Sorry if there are any sloppy mistakes or omissions. I've only spent about an hour on this post and have relied on studies I was already familiar with in my work as a labour and education economist and although I have read all of the studies at some point, not all of them were read forensically. I also have not proof read this, so some parts may be unclear.
What a thoughtful pushback to the too-cheery hype of educational reform! (And I say this as an educational reformer.)
>> "Schools... are designed to maximize motivation, but motivation is hard to maximize, and schools don’t do a great job."
I really appreciate the reviewer's identification of motivation as the bottleneck. Indeed, if the problem were anything else, then the spread of libraries should have led to the messianic age of education!
The reason I think massive improvement in education is possible is that there's a reservoir of motivation already inside schools that has not yet been systematically tapped: the content of the curriculum.
Math, science, history, and so on — down to spelling and, I don't know, small engine repair — are made up of innumerable small bits that have fascinated people in the past (that's how they got in the curriculum). The trouble is that they're pressed flat and delivered as unpalatable by the time they get to the classroom. The revivification of academic content is the best hope we have for an educational revolution that scales. (As a plus, it also doesn't require we figure out how to revolutionize a system that's been impervious to 100 years of reform. In any case, this is what I was trying to get at with my ACX book review in 2023.)
The reviewer comes across as a wise and careful thinker. If they're reading, I'd love to chat — feel free to reach out to me! (I won't compromise your anonymity.)
Who exactly was fascinated by those things in the past? Wasn't it a highly selected group, for any given topic? So this doesn't seem to tell us much about whether we can get the general population to engage with them.
Sharp point! But it also works the other way: we can look to see what aspects of the topic motivated them, so we're not working from zero. If this seems abstract, feel free to give an example of something in the K–12 curriculum that most people find dull, and I'll see if I can show how it can be re-humanized. (This is what I do for a living...)
I dispute your claim that schools are successfully bringing knowledge to even the low self-motivation students, because they often don't work to even teach in the first place. It's not just the "learning loss" like what Scott talks about where he forgot pretty much everything he doesn't use or see reinforced in pop culture; I've tutored multiple students in calculus classes who can't find the area of a rectangle. If their evaluations were working at all they wouldn't have been advanced to this class in the first place.
How is this possible? Imagine a teacher fails to teach anyone anything at all (standardized tests and personal experience both confirm this can happen). Students who fail would be held back a year, but holding back children would anger parents a lot, so they are sent to the next year anyway with a fake grade. Obviously, children are considered people with agency, so even if they are able to learn the subject elsewhere on their own they are still sent into the class, by the police if necessary, even if their teacher does nothing but scream into their faces and get really angry at people who ask clarifying questions, like Ms. Mandeville. (The actual name of one of my teachers!)
Most of the time schools aren't that obviously bad, I admit. Possibly most schools "merely" teach things that end up forgotten while making kids hate learning enough that they stop being curious about the topic if they started out curious. But the school system is so dysfunctional that it can't even detect if anyone learned anything at all.
This review rubs me the wrong way quite a lot; unfortunately, I can't give it the time it would deserve, so here is a short version. I will start with the parts that I agree with, to get it out of the way.
1) I fully agree that you can't "teach critical thinking" divorced from content. It is one of those things that sound great but when you look for substance behind the words, you won't find any. (Called "applause lights" in the Sequences.) How could someone think critically about e.g. World War 2, if that person had no idea when the war happened, who participated in it, and other similar encyclopedical knowledge. What would their opinions be based on, if not facts? A random edgy tweet, or a convincing (for someone who doesn't know any of the facts) article on a blog?
It actually works the other way round. You get some data in your head, and sometimes they connect, maybe in a way you did not expect. (And you cannot simply look them up on Google, because you do not know what will connect to what.)
I would even go further and say that, as far as I know, we do not have any evidence that "critical thinking" can be taught *at all*. I think there was an experiment with some university students who studied critical thinking, which showed that after passing the exams they were not any better at critical thinking compared to their classmates who didn't attend those lessons. (Also, "knowing about biases can hurt people" in the Sequences; we all probably know people who only learned the techniques of "critical thinking" superficially so that they can use them to reject inconvenient arguments.)
2) I agree about the self-reinforcing nature of motivation + study habits + skills/knowledge. (But I disagree that the third one is "intelligence".) Even getting slightly ahead of your peers can start the feedback loop where people praise you for your skills, you feel good about the topic in general, which makes you more willing to study it further, which further increases your skills, etc. And there is an equivalent opposite loop where falling slightly behind your peers makes it all feel hopeless, because you would have to work hard only to become average, which is no big success.
But from my perspective, this makes separating children at school by their physical age a horribly bad idea. That's how you create the losers! I mean, if someone is e.g. literally retarded, that is going to be obvious no matter what. But if someone is merely a few months behind their peers, the school system makes it super visible by contrasting them with their peers every day and giving them bad grades. (To compare with the life of adults, no one cares if you e.g. get your driving license one year later.)
3) Yes, it is easy to create an awesome school by having a random idea (anything that is not actively very harmful) and selecting only for good students. Good students create great learning environment, and you can then attribute that to your excellent idea, or your excellent teachers, or whatever. This can create a self-sustaining loop, when the school becomes famous, so many students apply, so the school can keep choosing the best ones.
This shouldn't be taken as a proof that good ideas are impossible. But it invites a lot of skepticism about ideas even after you have seen a successful school that follows them: maybe the true success is in the constellation of the students, not in the idea.
...okay, and now the remainign 90% that I disagree about...
For starters, I roll to disbelieve in the studies on increasing IQ. That doesn't even pass the smell test: IQ is defined *relatively* to people of the same age, with the average being 100. So if the school attendance is mandatory, and the average of the kids in the first grade is IQ 100, and the average of the kids in the second grade is IQ 100, and so on... where the hell is that 1-5 IQ points per year increase? Does not make sense, mathematically.
(Or do you mean that the kids in the second grade are smarter relatively to the kids in the first grade? Yeah, that's called "growing up"; I am sure it happens a lot outside the school, too.)
The repeated topic in this review is that almost all kids need some kind of external support / motivation / pressure / direction / structure / whatever we may choose to call it; and that although some kids only need a little or none, most kids need a lot. Without the system, the kids who need a lot of the external force would utterly fail. And... that's true... but it also seems irrelevant when comparing the existing school system to the alternatives, except for those that advocate zero use of external force, such as unschooling, some specifically American misinterpretations of constructivism, and the naive idea that if you give computer lessons to the kids you will never need to check whether they have actually turned on the computer and actually clicked on the lessons. But those are far from the only possible alternatives.
The original experiments with MOOCs failed, as far as I know, mostly because they were done in a way "here is the digital content, make sure to finish it before the end of the semester", and most students ignored it until one week before the end of the semester... and then it was too late to learn in before the exams. Yes, that specific approach to using MOOCs fails. But instead of rejecting the digital education completely, I think the proper lesson is to make sure that the students keep doing the lessons during the semester. Which, once you think about it, is trivial to implement: at the end of each week, check how many % of the semester have passed, how many % of content each student completed, send them an e-mail if they are behind, and send an e-mail also to their teacher is they are significantly behind.
These days, many MOOCs require you to proceed at a predetermined speed, which is horrible for an adult person who is not a student, doesn't have a predictable schedule, and just wanted to study something in their free time out of curiosity... but it addresses the problem for the students who use those MOOCs at school. There are often also forums where the virtual classmates can discuss the lessons and ask each other questions.
Similarly, the main problem with COVID-era remote education, in my opinion, was that people were totally unprepared for it. You had an improvized Zoom call where the kids didn't even hear the lesson properly, and... often, that was all. Yep, didn't work. But if you knew e.g. one year in advance that you will have to spend a few months teaching remotely, there is so much you could do better. For starters, you could prepare sheets with exercises, and deliver them by mail to students' homes. (Or, depending on the socioeconomic class, just send them some PDFs and tell them to print them.) You could prepare a nice video of the explanation, with clear image and sound, and upload it to YouTube. (If multiple teachers did this, you could also send your students links to other teachers explaining the same topic, so maybe they would benefit from having the same thing explained in several ways.) But more importantly, this could all be connected: watch a video... keep answering related questions on the paper... communicate in chat... and then do an automated exam. Maybe this wouldn't work perfectly, but I think it would work much better than mere improvised Zoom calls.
"In general, research has found that leveled reading reduces the achievement of the readers who struggle the most. We might intuitively think that reading an easier book would benefit students who have weaker reading skills, but that intuition seems to be wrong."
My intuition would suggest to give everyone an easier text first, more complicated text later, after they can read the easier text correctly. Where can I go collect my Nobel prize for education?
As a sidenote, giving up at the first failure seems like a repeated pattern in debates about education. Imagine someone talking about e.g. cooking the same way: "Once I tried to cook a soup. It was too salty. The idea of 'cooking a soup' is debunked. Anyone who suggests that you can cook a soup at home is clearly wrong, they should read my report instead." I find it ironic that the school system is so bad at learning. Then again, school has practically redefined "learning" to mean "memorizing the words of the authority", so perhaps the idea of learning by experimenting is at a blind spot of most experts on the school system.
Another sidenote, even the words "no-structure", "low-structure", "high-structure" rub me the wrong way, because they are so... structure-centric. It's like calling a bird outside the cage a "no-cage bird", which kinda suggests that the cage is the bird's natural habitat, and it's the absence of the cage that makes this specific bird weird. I would prefer an opposite perspective, e.g. to call the students high-agency, low-agency, and no-agency... or some word other than agency, but the focus should be on the quality that makes learning happen (and which, when low or missing, can be substituted by the structure).
"incremental rehearsal is a highly-structured way to teach students multiplication facts."
Well, that kinda sounds like something a computer could help with, just saying.
[...CONTINUED...]
"The no-structure learners will always be bored, as long as we are committed to putting them into classrooms where everyone learns the same thing."
Wrong. There is a vast difference between "learning the same thing" and "learning the thing that you already know... at an excruciatingly slow speed... every day... for years". Both can be boring, but it's not the same quality; given the choice, I would choose the first one in a heartbeat.
Also, maybe we shouldn't teach the boring things, or we shouldn't teach them the boring way? Why are we even doing that? Yeah, because that's the only way that works well with the high-structure students. Basically, the students who resist learning require an approach that is almost the *opposite* of someone who wants to learn. It's like, for the student who wants to learn, you build the track; for the student who doesn't want to learn, you build the barriers along the track so that he can't escape.
"[Kids at gifted school] would be doing more advanced work than peers in regular schools, absolutely. But often only by a year or two."
Well, a year or two is a lot. Imagine coming to the university two years earlier. For example, if a person wants to spend a fixed number of years at school, how much more they could learn if they could replace two years at elementary school listening to the stuff they already know, with two extra years of university education? The same cost for the educational system; much better outcome.
From my perspective, the problem with many gifted schools is that they focus too much on intelligence, and fail to notice things such as love of learning. If you combine intelligence with love of learning, you get magic. If you combine intelligence with the lack of love of learning, you get bored students sitting in the class for gifted kids.
Note that the love of learning is *not* the same as "no-structure". Those are two different dimensions. If you make a voluntary class, the people who will come are those who love to learn, and prefer to have some structure for that. I know many people like that.
"The core challenge of compulsory public education is motivation. The best solution we’ve found is to send kids to school beginning at age 5 (or earlier if we can), before they can reliably form long-term episodic memories."
Are you insane?! School is typically the thing that *kills* motivation in children. Small children are curious about everything, they can't stop asking question. It is school that makes them associate learning with boredom. It is school that associates learning with punishment (even if merely verbal and social).
"The system motivates high-structure learners to keep up with their peers, though that motivation does gradually fade over time."
So we basically sacrifice motivation of some kids to create motivation for other kids... and even that doesn't work well in long term. Cool!
"Maybe the best way education can contribute to all of these causes is to focus on providing a broad, basic education to as many children as possible, and maybe our current system is the best idea we’ve had so far for doing so."
Maybe the current system is the best for those children who don't have a hope to achieve anything beyond the basic education. And maybe those children are a majority of the population, like maybe even 80% of it, although I doubt it.
It would still make sense to have a separate track for the remaining 20%. And high school is too late.
As a former teacher, it was my experience that the class is often slowed down by one or two kids who are either clearly uncooperative, or somehow damaged. Maybe those schools were not representative for the population at large, and the actual ratio is different at other places; I don't know. Sometimes, if such child became sick for a few weeks, the classroom experience changed dramatically in their absence... and then immediately reverted back. As a result, sometimes 18 out of 20 students cannot get the education they might get. And this is considered perfectly acceptable. It is democratic, inclusive, insert your favorite buzzword... Yes, you can get a dramatically better school by simply *not* admitting kids like this, and then people complain that this is cheating. From my perspective, it is sanity. Instead of complaining that it is unfair when only a few schools do it, all schools should be doing it. (That is, except for a few selected schools that would be specialized for kids like this.)
Basically, we are trying to make the school same for everyone, and then it barely serves the lowest common denominator. All improvements will be criticized that they only work for a certain subset of students. But from my perspective, that's okay. The students who can get a better system, should get a better system; different systems for different students. And the students who couldn't work in a system different from the current one, should stay in the current one. But they should not be allowed to drag everyone else down.
Without making any claims about whether any particular teaching technique works, it seems pretty clear to me that "critical thinking divorced from content" is a coherent concept and in principle is learnable. Our esteemed host made the case for it better than I could, nearly a decade ago (https://slatestarcodex.com/2015/08/04/contra-hallquist-on-scientific-rationality/):
"If we want to get all hypothetical, we can imagine some kind of theorizing contest between a totally irrational person with an encyclopaedic knowledge of cell biology, versus a very rational person who knows nothing at all about the subject. Who would win? Well, who cares? Whoever wins, *we lose*. We lose because I want the people working on curing cancer to be good at both cell biology *and* thinking clearly, to know both the parts of science specific to their own field and the parts of science that are as broad as Thought itself. I have *seen* what happens when people know everything about cell biology and nothing about rationality. You get AMRI Nutrigenomics [1], where a bunch of people with PhDs and MDs give a breathtakingly beautiful analysis of the complexities of the methylation cycle, then use it to prove that vaccines cause autism. By all means, know as much about methylation as they do! But you’ve also got to have something they’re missing!"
[1] http://www.heartfixer.com/AMRI-Nutrigenomics.htm
This only addresses one piece of the problem but -- what if in order to vote people had to demonstrate decent knowledge of structure of government, the last 100 years of history, laws, political parties, taxes, treaties, voting, and let's say 3 current issues? AI can generate lots of tests and ensure that they are of equivalent difficulty, and score them, both multiple choice items and short answer.
Sorry, what problem does this solve? The post isn't about governance quality.
Oh sorry I thought that every time the word "school" appeared in the review the writer meant "Government." Like that "school" was just a slang term. I misunderstood the whole review! Oh dear I am such a silly.
Is your point that if we disenfranchised the ignorant then we could just give up on educating unmotivated students, because their ignorance can't hurt their fellow citizens? I do not think that that follows. In a highly complex information-age society, externalities are everywhere, not just in electoral politics. You want the people you transact with in the marketplace, or who live next door to you or pass you on the street, to have a baseline understanding of how the world works and of our shared culture; it doesn't always matter, but sometimes it does. And being uneducated puts you yourself at a disadvantage as well, as it makes it harder to tell when you're being duped. Yes, there's a lot of waste and uselessness in the K-12 curriculum, both for unmotivated students and for motivated ones, but this does not move me to give up entirely on the concept of a universal baseline education, albeit maybe a smaller one than we presently have.
Also, disenfranchising a large fraction of the population would mean that elected officials would be free to entirely disregard their interests, and that seems *extremely* bad and I would have to see a damned strong case that it was worth the cost.
%s/democracy/communism/g
Importantly DK is assuming an ultimate point to school, inside the perceptive point that it's about "motivation." Motivation for what; learning; this elides the point.
Consider different functions of schools:
1) Teach kids useful information that will make them more effective and productive adults.
2) Day care. Get kids out of the way so the adults can work.
2a) Divert kids away from worse activities like watching TV all day, drug use, unprotected sex, crime.
3) Produce a ranking of the kids relative to each other.
4) Stealth social services. A square meal a day, and put kids in contact with some caring adults.
5) Produce entertainment: school plays, band, quiz bowls, beauty pageants, sporting events.
6) Wealth transfer from prosperous blue areas to unproductive red areas via teacher, administrator and staff salaries and spending on school infrastructure.
7) Political indoctrination, e.g. the Revolutionary War is an origin story for the US meant to create political cohesion.
I agree with DK's cogent point well-made that school is structured for motivation across the whole population. But the "motivation" is for "perpetuation of school". Schools have evolved to perpetuate schools, unsurprisingly.
Upsettingly for people who like to envision better schools, not going to school turns out to be quite bad. Smartphones in schools are undoing a chief purpose of modern schools, to stop kids from watching TV all day. Smartphones have put TV right into schools. Movements to ban phones in school are very good.
The Federal Reserve, of all people, did a careful study of how kids spend their non-school time, and found that diverting kids from bad activities like watching TV was important:
https://www.federalreserve.gov/econres/notes/feds-notes/how-do-children-spend-their-time-time-use-and-skill-development-in-the-psid-20200526.html
The idea of “high structure kids” and “low structure kids” is important to a lot of the author's argument. How established and validated are these categories?
There are some dimensions on which almost everybody falls into one of a few categories: gender, blood type, eye color, handedness. But on most dimensions people are distributed in some approximation of a bell curve: most people are in the middle, and a considerably smaller number are quite a bit above or below the middle range. Some in the absence of evidence to the contrary, I'd expect that most kids are medium-structure.
And then, on the dimensions on which variance takes the bell shape, a lot of them turn out to be situation-specific. For instance, consider aggressiveness. If you look at most people across a variety of situations, you will see a lot of variation in how aggressive they are. They might be very aggressive chess players, driving grimly towards the endgame, but very casual when playing tennis, chatting about this and that while they play. They might push very hard in work discussions when certain issues come up that they care a lot about, but shrug and say “whatever” about other issues. They might be very irritable with other drivers, but patient and gentle with their teenage daughter even when she’s being quite difficult. In fact there are not many “qualities” at all on which people's scores are fairly stable across different situations. Intelligence is pretty special because it is one of those qualities. There is a lot less variation across situations: People able to remember and recite a long string of digits also tend to have a llarge vocabularies, and to be good at mental math and at reasoning, and to have la large store of general knowledge. So in the absence of evidence that how structure-dependent somebody is stays the same across settings and situations, I'd expect that most kids vary quite a lot.
So unless there’s some pretty solid evidence that most kids are either high structure or low structure, and that they stay in whatever category they’re in across a lot of situations, then a lot of the author's train of thought about schooling and what works is unfounded. Is there in fact some well-validated psychometric measure of high/low structure, one that shows the distribution tends towards being bimodal and that kids’ high- or low-structure membership tends to stay the same across school subjects?
I’m told one-room schools were even better than grouping by age, because you could hear the lessons from higher grades before you got there. My father’s family was from a tiny rural fishing village. When he moved to California at 16, they were years ahead of their piers in education.
Here’s an article discussing some of these advantages: https://www.forbes.com/sites/brandonbusteed/2020/11/07/why-the-one-room-schoolhouse-is-a-vision-for-the-future-not-just-a-relic-of-the-past/
I’m sure it didn’t work for everyone and it’d be hard to implement at mass today. But maybe there is inspiration there.