I liked a lot of this. But whenever I read a long thing that attempts to dig into education research and then cites Bloom's 2 sigma result without qualification, it immediately changes my confidence in the analysis.
This review of Bloom's two-sigma claims by Paul T. von Hippel would itself probably earn a top spot in the ACX review contest. Worth reading. Thanks for posting.
The claim about two-sigma tutoring effect was based on a 3-week study by Bloom's two PhD students.
"... they weren’t obtained by tutoring alone. Instead, Anania and Burke mixed a potent cocktail of interventions that included tutoring; training and coaching in effective instructional practices; extra time; and frequent testing, feedback, and retesting....Bloom’s two-sigma claim had some basis in fact, but it also contained elements of fiction."
“Tutored students received extra testing and feedback. Burke’s and Anania’s two-sigma intervention did involve tutoring, but it also had other features. Perhaps the most important was that tutored students received extra testing and feedback. At the end of each unit, all students took a quiz, but any tutored student who scored below 80 percent (in Anania’s study) or 90 percent (in Burke’s) received feedback and correction on concepts that they had missed. Then the tutored students took a second quiz with new questions—a quiz that students in the whole-class condition never received. If the tutored students still scored below 80 or 90 percent, they got more feedback and another quiz.”
In Made To Stick the authors drive home one of the most important tools for improved performance is regular testing (ie students with 9 quizzes/year outperform students with a midterm and a final on almost all metrics). It seems like Bloom’s tutoring was combined with more regular testing than the control group. I expect the impact he found was a conflation of the combination of (at least) both methods
I don't see how that really negates the point about 1-on-1 instruction accelerating educational performance, though? The point is that a single teacher focused on a single student can give rapid, tailored feedback and quizes, they can't do that for an entire class nearly as easily.
The article corb linked to did not discredit 1:1 tutoring. It just said that Blooms original research was overstated and that a 2-sigma improvement has not been replicated. The article says that stand alone 1:1 tutoring only gets you about a 0.4 or 0.7 SD improvement (still a lot!)
So it’s still very effective and possibly even more effective when combined with regular testing for comprehension.
Main takeaway stays the same, but just caveats on Blooms original claims
Are you saying that the overall cocktail of interventions that Bloom implemented has not replicated the same benefits, or that tutoring alone doesn't deliver the ~2SD increase? Does the overall cocktail of interventions require 2 PhDs per student working 15 hours a day, or something?
I recommend reading the book on expertise that this review linked. Having rapid feedback on performance is an absolutely essential part of the "deliberate practice" mentioned here. It's not at all surprising that more frequent feedback works better (although I would still think of something that happens 9 times a year, which is around once per month in a standard school year, as a "unit test" rather than a "quiz").
I don’t love how qualifying Bloom’s 2 sigma has become a shibboleth for being able to talk about education research. But I agree the study has issues, has regressed, etc.
I would say that we hold it as an engineering goal, not a research goal. We believe we can get effects that large and are trying to do so. (I work for this org.)
Most of that criticism, as I read it, isn't relevant to the way the 2-sigma problem is being used here. If you are citing the 2-sigma study to promote after school tutoring, or an AI tutor, or some other tutor-specific intervention, then the details in there matter a lot.
But if you are citing it as a way of saying "we know ways of dramatically increasing performance, but they are too difficult to scale" (as it is being used here), then it mostly doesn't actually matter that they did more than just tutoring, or that it was a special kind of tutoring, etc. If anything that strengthens the point, that those results are _even_ harder and more expensive to scale.
The fact that it was on a narrow set of tests is somewhat more convincing, but, at least for me, doesn't really do much to change the larger point. I think adding those caveats, in this context specifically, would have just been extra cruft (on an already long essay) that wouldn't have changed the message at all.
I would love to hear a perspective from another parent from the acx community to see if their parenting ties out. It strikes me as almost too good to be true.
Yeah… I’ve seen this. The commenters are all Reddit-flavored criticism from people who have no direct experience except for one (“dipnoob”), who said his daughter didn’t like it after going for a year. In his post history he also says he has an 18 year old trans daughter, and given the schools association with Elon musk it seems like that could easily be an area of conflict that isn’t necessarily related to education quality for the average student.
There are also two comments from actual students at the first link (ProgressPlane4465 and Relevant_Habit6774), both have no other reddit history which may or may not be suspicious (I guess it's easy to sign up just to comment on a single post and then it gives you this sort of random username? But also someone who wanted to astroturf would just do the same)
> I know a family who goes to Alpha. They don’t learn history, and the other things I’ve gathered from the family about the school make me certain I would never send my kid there.
Its from 7 months ago so I wonder if they have talked to this family since 2022.
Hi Nicholas! I have had 2 students at Alpha's flagship location here in south Austin these past 3 years and have been blown away by their experience. Not usually active in ACX comments but have been reading for years and my husband is a prominent writer in this social neighborhood broadly (Byrne Hobart of The Diff, https://x.com/ByrneHobart/ )
My 9yo daughter is quite high ability and I very much doubt that anywhere else would have let her move quickly. I never had to make a single request for accelerated material or schedule a single conference to complain that school was too easy or boring for her, I have basically no friends in Austin who have not faced these issues in various other charter and private schools for their children. She has maintained the 99th ability not just for NWEA MAP score achievement but also the top few percentiles for growth throughout these 3 years.
My 7yo daughter is also bright but a tough cookie, the label is "2e/twice exceptional" i.e. on the spectrum but also gifted. She spends very little time on the apps because the school environment is challenging for her. Her growth has also nonetheless been incredible - I am extremely confident that the academic environment Alpha provides via its 2 Hour Learning platform is the most potent available today. https://x.com/gtmom/status/1931058079745056974
I believe in the model so much I've gone to work for Alpha/GT school and am moving my daughters to the GT school this fall.
I am a parent at GT School, the gifted school of Alpha. My son, Nate is reading at a 7th grade level, is in Pre-Algebra, and similarly ahead in science and language. He is 10 years old, and a young 10 with his birthday mid-April. After being bullied in two different (well-renowned) public ISDs we chose to switch to private. His first week of school he was beaming and exclaiming that kids “got him” at this school. Alpha truly does foster a love of learning and does accelerate learning. Though I agree it may not be for everyone, my son loves it and is thriving.
Seems like most homeschoolers are already spending way less time than they would in a traditional public school, so this piece overstates the impressiveness of Alpha's homeschool pilot results.
Second this. It’s nice that it’s preassembled, but my kids are homeschooled and advancing at 1x speed on less than 2 hours a day (for a lot less than $10k per year).
I always point out that Finland and South Korea get near-identical PISA scores despite vast differences in hours worked, and Scott already did a piece arguing that homeschooled kids who do little or no work are only about a year behind their public-school peers.
That said, learning 2-3x faster with only 2 hours a day is pretty impressive, for the full Alpha program. It's a shame the entire progressive establishment are impervious to evidence.
I haven't read the entire review (yet) but 2-3x sustained would have the kids performing at college freshman level by about grade 5. Do they? Or does the 2-3x mean something else?
I think the article mentions top-end students performing at the level of high-school graduates by grade 3, so... conceivably, yes? Maybe you hit a point of diminishing returns at some stage once you get into adolescence/adulthood- there are a lot of interventions that give impressive results in kids that don't persist into adulthood, although some socialising interventions do seem to be longer-lasting (I think Charles Murray wrote about this at some point.)
But even if the only permanent benefit is *not* squandering vast amounts of childrens' time that's well worth pursuing. Bring back child labour, unironically.
So this is mainly measured by MAP test scores, which heavily correlate to SAT/ACT scores. This is mainly going to measure reading and math proficiency. So my guess would be, yea, a fifth grader in this system would probably be able to read and understand Fagles’ Iliad (though the depth of understanding is going to be limited by emotional development). Having the reading ability would enable them to have at least a conceptual understanding of Newtonian physics if they’ve encountered it, and I would think that if the curriculum contains instruction on application of Newtonian physics there’s no reason they can’t understand that as well.
It's in the review. It's entirely MAP score based. So top level 3rd graders get the same MAP score (multiple choice testing) as median high schoolers. So they probably don't understand Newtonian physics or read the Illiad....because median high schoolers can't do those things either.
(I work for Alpha.) We don’t think our homeschooling platform is impressive (yet). And yeah, “you can use time more efficiently than schools for learning school things” is not magic or unique to us.
It’s still true though, and I’m happy to be working on a project that elevates it to a constraint in the school system.
This right here. Homeschoolers often spend way less time per day and get good results, with a wide spectrum of curriculum and approaches. That these Alpha folks did something similar and got results like that should not be news.
This is really interesting and will directly change my parenting from this day forward: my husband and I spend 30 minutes every morning 1-1 tutoring our kids, and I will start adding financial incentives from next week. Our 3yo can read and write (not very well! But better than most 3yo!), and the 6 yo is very good at reading but not quite at adult level fluency yet. I'll be paying him to read the Roald Dahl corpus and for completing handwriting worksheets. The 3 yo can get money for finishing each phonics book. Thank you for an informative review!
Any tips on introducing early reading and writing? I have a 2.5 year old and I’m very encouraged by these non traditional models. Big fan of 1:1 tutoring. Thanks!
There are two approaches I've used. The first is Siegfried Englemann's famous "Teach your child to read in 100 easy lessons". The lessons are not easy and it took us 400 days to complete the whole book with my first son. However, it was successful and my son did learn to read. It was a painful process and the Distar orthography was somewhat frustrating and at odds with the Phonics approach taught in his reception class. Fortunately we live in England which has fully adopted evidence-based reading education. I didn't do the writing element which I regretted - my son's handwriting and writing ability was far behind his reading level which held him back somewhat. I will be teaching him "joined up" handwriting this summer before Year 2.
With my younger son we decided to use a different approach. We bought Read Write Inc flashcards (the phonics scheme his eventual school uses) and taught him the Set 1 sounds. We started "blending" very early, but this is the part that took a huge amount of time and perseverance for both my sons to be good at. This is going from c-a-t to "cat". I.e. blending "c" and "a" together to get "ca". We worked for 3 months on this before he got it - we basically did direct instruction (e.g. going "c-a, ca" now you try - the copying works!).
We then bought a different phonics scheme books, so that when he starts Reception (this September - he'll be a summer born kid - we're not holding him back - I still question our decision) he won't be bored/remembering the books so they can properly assess his level. He started reading these - exceptionally painfully - but it worked. I then introduced writing, again using RWI - first using wipe away sheets, then buying the RWI workbooks. I also make him write his name every day - he's a pro at that now. He writes something every day (i.e. The cat is is on the mat).
We've now moved onto Set 2/Set 3 sounds. And we've memorised high frequency non-decodable words, as well as working on sounding things out in your head.
It's pretty painful at times but very rewarding to see his progress.
How do you prevent them from just memorizing the phonics books, and instead actually learn to sound them out? I feel like working through sounding out with my 4 year old, she'll struggle through it to get a word, but then she just kind of knows it, and doesn't have to sound it out so much the second time and I feel like she's not actually getting good practice with reading but just partial memorization of what the word looks like.
You do both. We have "speedy green words" which we get him to sound out in his head and just "say" and harder words he has to sound out. You sound like your kid is already getting a great level of semantic knowledge where they don't need to sound i tout - lucky you - mine laboriously sounds out every word (the younger one) apart from the non decodable memorised words.
Yeah I know it's sort of a good thing, but we can't use the same texts too many times to practice sounding out so she's acquiring words but not getting that much better at sounding things out. I guess it will come, and we just have to get new books a lot.
OK have reread this. We don't have this issue as the phonics scheme we are using has a huge number of books. We reread each book twice and then move on. There are over 100 decodable books in the scheme.
When my wife and I were teaching our child to read we had "hold out" books (maybe a Clifford book or a Little Critter book) that we knew the child had never read before. When the child could read one of these, we'd score his as reading.
This did NOT prevent him from simply memorizing all the words that would appear in the book from exposure in other books, but that is what we did. The "if he can read a brand new book" metric seemed to track what we meant by reading.
IME it's mostly genetic. My first learned all his letters by 18 months. My second still didn't know them by 5y. I didn't do anything differently; in fact I did so much more with the second because I was concerned. Turns out the first one was autistic and hyperlexia is a symptom. More of a curse than a blessing, in the end. Meanwhile the second did eventually catch up.
I was convinced by this review's defense of incentives, but I *don't* think paying a kid to read Roald Dahl books in particular is a good idea. Reading fiction is something you want kids to enjoy for its own sake, and I think there *is* convincing research supporting the view that rewards undermine that sort of intrinsic motivation. (See https://www.alfiekohn.org/article/newt-gingrichs-reading-plan/ for a brief overview.)
I think rewarding kids for completing educational modules is different, because we don't really care whether kids enjoy those things for their own sake--the educational modules are obviously just a means to the end of acquiring the necessary skills and knowledge. But if you want the kid to WANT to enjoy reading for its own sake as an adult, paying him to do it now is counterproductive.
Isn't the whole point that you give them £1 to read Roald Dahl books and it doesn't impact on their enjoyment at all - instead it may actually improve their motivation to read even without monetary incentive!
I can't see how paying my son to read a great book would make him less likely to enjoy it.
I also don't particularly care about intrinsic motivation. I don't see why anyone should - I'm familiar with Alfie Kohn and think his influence on education is malign.
I think in a perfect world, kids would read a few great children’s books and spend their free time doing other things than reading.
But my experience of kids myself included and, coincidentally, my genetically related kid - who go on to become constant readers as adolescents and adults, is the personal search for books, almost furtive and almost entirely undisturbed; a lot of which is plowing through a fair amount of garbage, or short of that at the very least, deeply unedifying books.
I spent my whole childhood reading - it was an immense source of pleasure for me. I hope it will be for my kids too. But why on earth shouldn't we try to make readers, given the links between volume of reading and educational outcomes?
I guess we need Scott A. to craft a survey question to get at: if you are a reader - someone who, at the thought of a threat to vision, as is beginning to happen at my age, of course I’m nearsighted and astigmatic but I don’t mean that - your second thought is not: to lose the sight of beautiful waterfalls, or relatives - but rather: the fear of not being able to read text many hours a day!! - if you are that …
Were you made by your parents? Would you not be so without parental involvement? Did your parents have anything to do with this at all, or was it more like a pulling away from them into a private world?
I don’t know many readers (any beyond the nuclear family?) so they may have been generated by those contests in school or parents saying, now is your 30
minutes you must spend reading.
Some of us did indeed get in trouble for reading too late; and I myself once took my son’s copies of LoTR and hid them away.
I developed no useful skills as a kid but it may not be fair to blame reading. I watched a good deal of TV as well - enough to be full up for my whole life as it turned out.
This is a very interesting take I've never seen before - what's the negative of reading? That many books are bad? What would you want kids to do instead?
I can see why common-sensically it might sound silly that paying a kid to read would make them less likely to enjoy it, but there does seem to be a fair amount of evidence that it can genuinely have that effect.
For example, there are experiments that show that if you reward kids for doing a particular activity on one day, they're less likely to freely choose to do that activity if given the option on the next day. Now, admittedly, we have to take basically all psychology studies with a grain of salt, but in this case, I think this really does line up with my own experience of my own motivation. If I got paid to do something before, but then later I'm merely offered the opportunity to do it without pay, I have some tendency to now think, "Why should I do it if I'm not getting paid?" Doing it "for free" now strikes you as a waste of time. And this does seem to be born out by research.
Sending the message that reading Roald Dahl is something you need to be bribed to do, rather than a delightful thing you *get* to do, seems like a bad idea to me.
But if you don't particularly care about intrinsic motivation, I guess we might stand too far apart to have common ground for conversation. I can't imagine not caring about intrinsic motivation. To me, it seems that happiness in life absolutely depends on being able to enjoy for their own sake as many of the things you *have* to do as possible.
I think all the evidence we have shows that intrinsic motivation is one of the most heritable things there is! So to the extent my children do or don't have it, that's not something I can influence.
The research quoted in the article seems at odds with what you're saying - that actually rewards don't make kids less likely to pick that activity - they make them more likely to choose it!
There’s a big difference between an incentive to get over a hump and kickstart a process, and an incentive as a crutch to solve for an exercise that a child really and enduringly does not want to do.
If my picky toddler won’t try a donut (a real example), I can bribe them to take one bite, and then the issue is obviated. This kind of thing comes up all the time. The point is not to be an imperialist or a reductionist about incentives, it’s to have them in the toolkit.
My husband gives our kids money for household chores (which I disagreed with - I felt it should be an expectation not something to he rewarded). He's read my comments, disagreed with bribing to read books and has countered with incentives to use family Duolingo which I have agreed with.
Not sure why that should sound silly. After all it’s a pretty common observation among adults who make a beloved hobby their career is that they lose a beloved hobby.
Of course mileage can and does vary there – might also be an interesting thing to study: What about personality and activity structure makes adding material incentives to passion work or fail?
I say, don't micromanage. There aren't that many Dahl books. Let them read what they want and pay up. As a kid I hated when something that sounded fun turned out to have all these terms and conditions. "Good practice for being an adult", yes. "Incentive", no.
The point of the argument here is that external incentives is the first step on the ladder to internal motivation.
The hypothetical for reading fiction would be:
Child doesn't like (or maybe just doesn't care about) reading -> incentivized to read -> finds out that some books were actually enjoyable -> decides to keep reading on their own
That, to me, is at least plausible, as a way of encouraging a child to find out whether or not they might enjoy reading. For it to work, you probably want to be incentivizing a wide range of different books until they find out what they like.
There's also the skill hump to get over. When you're not yet that good at reading, reading a whole short book is a lot of effort, and the story isn't enjoyable because the struggle of actually reading it is so hard. After a lot of practice, you get better at reading and it becomes easy, and the story is its own reward.
I'm still undecided on the idea of monetary rewards for basic tasks, though. I know my kids will do just about anything for a dollar, but I also know that if I overuse this then soon they'll be refusing to do anything without getting paid... and then they'll start negotiating for more money.
Curious: do you think *monetary* rewards are the best way to go with a 3yo? Back when I was bribing a 3yo to read, it earned him ice cream; I'm curious about the relative merits of ice cream vs cash. (That former 3yo is now 6, and I'm planning to bribe more reading with movie nights -- TBD how well that works.)
I think there is some evidence bribing with food might lead to a bad association with food = rewards = obesity. I don't know how right that is, but in any case, cash rules all and you can use cash to buy what you want, including an ice cream. I think he's probably too small to get it just yet, but we'll see.
My approach was "the 3yo should get an immediate reward; an extra dessert seems like a sustainable level of immediate rewarding". (This is presumably the mechanism behind stickers, I just never quite figured out what kids are supposed to *do* with the stickers they get, and consequently my kids don't quite seem to know either.) If cash works as an immediate reward for a 3yo (and if you actually keep cash in small denominations on hand! -- part of my problem was that I don't), then I don't disagree that it's equally good.
This is really interesting, thank you ! As a parent of similarly aged kids, do you have any advice on how to make those 30 mn a day a pleasant time ? I feel that every time I try to teach my son, I have a hard time staying calm and he is not very interested either.
Stay calm, over praise your child for tiny successes ("oh wow!!!! you're soooo clever - that's right - that sound is "oy, a toy to enjoy"!). Be very firm; the exercise/reading does not end until you say it ends. Use rewards/incentives - we already say "you can watch TV, but first, finish your reading". Be very clear at the beginning what the expectation is - i.e. "first we're going to practice your writing with the yellow sheet, then we'll do your red book, then you'll write your name, then we'll practice some sounds, then we'll do red words, then we'll read two pages of your book".
I think the home school product has become available. But my googling can’t find it. It’s been quoted at $10,000/year though and doesn’t include afternoon programming, “babysitting” or incentives
In Alpha's pilot of this program, however, it has been much harder to get the 2x+ learning results in the home environment without the wonderful guides and various tailored incentives.
Oh certainly, it's just that homeschool users (even early adopters) may not understand how critical the motivational model/incentives really are. We can probably do more in that regard to educate/support as we continue to experiment and roll it out, but ultimately Alpha cannot control whatever is happening not within its walls.
I homeschooled another child of mine for a while and it did not go well for many reasons but I did find that incentives at home felt much more manipulative to him - whereas learning alongside a bunch of other kids who are also racking up their app time and cheering each other on and sometimes going on outings together as a reward feels way more natural. At home with just me and him it felt admittedly lame : "do your work!! don't you want these stickers?? chop chop " 🥴
(Alpha parent who now works for them) There is AI behind the scenes refining our apps platform, including eye tracking and lots of interpretation of student learning behavior to determine when the material needs revision vs. when the students are choosing to waste time, etc.
But you're correct that the "AI tutor" is not like talking to an LLM/chatbot or anything of the sort, and that many people who see various media coverage are misinterpreting it as such.
I wasn't part of that strategic decision, to lean into "AI" branding but, as a parent who was considering Alpha a few years ago, the "AI" component didn't loom large in my own decision making. It has probably changed by now as AI enters the popular discourse more and more.
Given the section about their current consumer base, it might be that it's really unpopular among the general populace (although even that isn't actually obvious to me) and much more popular/less unpopular among the group of people who are considering these kinds of schools.
If that's the case, it might make sense now, but it could wind up being a poor choice down the road depending on how large their scaling ambitions are.
I believe it's for PR & virality purposes. Everyone's wondering how AI will "transform education" and journalists are hungry to write about it. Alpha Schools has existed for a while, has always used personalized apps, and only got this level of media traction when they started saying it was 'replacing teachers with AI.'
If that makes some parents and educators angry, all the better for provoking discussion.
I mean, there’s also a sense in which it’s true. The in person teachers don’t give academic instruction or do lesson planning. AI is integrated into evaluations of mastery and decisions about next steps.
But yeah, I do think that people are (naturally and understandably) misreading it as “replacing teachers with chatbots”.
Hi Matt! Definitely agree there's a sense in which it's true - though the guides give some form of instruction during afternoon workshops, yes? And their presence (paired with incentives) seems to meaningfully drive completion of the 2 hour learning, given that homeschool students achieve less impressive results. And the virtual coaches available for extra support when kids get stuck could also somewhat qualify as teachers, right?
And yes, although I'm pretty deep in this stuff (spoke to people at Alpha back in 2020 before Gen AI boom) I was surprised to read that chatbots aren't a bigger part of the 2-hour-learning model now.
I think when people hear about AI replacing teachers, they expect this to mean that education is being done cheaper and more scalably - whereas Alpha's pitch is AI = education done faster + more effectively. It's more like "AI replacing some of the typical functions of a teacher's role, while actually having a student:teacher ratio lower than typical schools and paying teachers more, leading to a counterintuitively higher tuition price." Of course this is addressed really well in the piece, but I agree there are major scaling concerns. Nonetheless I do think there are ways to make the model work to some degree at home + virtually.
Gonna try to find time to write a higher-level comment on the pedagogy since this review is so rich and fascinating. For full disclosure in this thread, I've spent the past 5 years building Prisma, a virtual school in the same innovative education space.
An enormous portion of our costs right now are paying for AI models, which will come down.
Teacher salary is actually not the bottleneck, which I think should remain high. Nor is teacher:student ratio, which will be and in some cases is higher in other models that use the platform successfully.
The system does not function without teachers and the purpose of the system is not to replace teachers. The purpose of AI is specifically perfect and automate academics, which is a process that is not complete for some of the reasons you note.
Edit: I think the reality is that we are just dipping our toes into scale. There’s a lot to figure out. That’s the attitude from this side of the project.
Love all that! And I think that messaging will resonate and not offend - most parents know and feel that the best teachers are the best at driving relationships that enhance learning, and academics can be optimized other ways.
Hi Emily! I am Alpha parent who now works with them/Matt as well. You're right that there are several sliders here, not far from the usual "good, fast, or cheap" pick two situation. Unfortunately conventional education actually picks 0.
The counterintuitively high tuition price has several causes: huge expense of the behind-the-scenes tech stuff (building and improving the apps, this is AI powered with eye tracking data etc), guides who earn 2-3x public school teachers in the same areas with dramatically higher teacher:student ratio, and some of the workshops are quite expensive (indoor climbing 3 days/week, swim lessons, field trips to a ranch...). These could be adjusted in various ways in the future , like well paid guides in different ratio, cost effective workshops, cheaper buildings in lower cost zip codes.
The guides do create and implement the afternoon workshops, and some of those have used LLM/chatbot type AI, as a kind of educational dessert and not the main course. Here is a video about the recent "AI Imagineers" workshop, my daughter is the one making animal trading cards. https://www.youtube.com/watch?v=foGBJHdjhZg
This sounds very cool, though I neither live in Austin nor have a spouse who is open to homeschooling, so I don't think I'll ever be able to use it. Alas.
The one thing I *wouldn't* be happy with is the lack of a few things that I considered really essential to my own education: (a) significant periods of sustained silent reading for pleasure in elementary school, (b) in middle/high school, classes where everyone reads a classic work of fiction over a few weeks, discussing it as a group as they go, and (c) lots of practice writing, including creative writing, where it gets read and critiqued by a human--at least by a human teacher, but sometimes also by fellow students.
These are things that you can't (or certainly shouldn't) get AI to grade, where performance isn't going to be clearly quantifiable, but they should be included.
I had similar thoughts! I’d heard that kids in general aren’t reading full books in school these days, so if there were a way to combine that with the accelerated study in other areas, that would be ideal…
I went to a Montessori school from preschool through 6th grade, so I have a lot of appreciation for parts of this approach (I got to have a much more self-paced experience than most kids do, though obviously not to this extent).
The things I would probably not have liked:
-So, if you’re in the GT program, you have to do chess *and* quizbowl *and* public speaking? Sounds great if you’re into all of those things; not if you’re interested in some but not others. Or is the idea “some schools choose to include Latin or programming in their Official Curriculum; we’re choosing to include public speaking in ours, so everyone has to do that”?
-Similarly, I have this vision of my 7-year-old self cruising through 4th- or 5th-grade math and English while still officially not having “leveled up” from first grade due to not doing, say, the paper airplane thing (maybe due to some difficulty at first, subsequently out of something like spite).
-Not a fan of the idea of everybody learning just enough piano to play a grade 1 solo, and then not playing piano anymore. Maybe this is just my personal bias (for me, chess is the thing I would have chosen to pick up for a few weeks and then put down, but for them, it’s music).
Hi there, I am Alpha parent who's gone to work for them, my 2 daughters are moving to the GT version this fall.
Since core academic learning is so efficient when done on the Alpha apps platform, there is a ton of time for workshops, even with a relatively short school day that starts late (~8:45a-3:30pm). You're right that some kids will like some of those activities more than others, but they've got to be doing something constructive with the time, and the workshops rotate each short session (~6 weeks) so it's not like being stuck coming into school at 6am every day for quiz bowl that you hate.
I have exactly that kind of salty first grader lol (cf https://x.com/gtmom/status/1931058079745056974 ) . In practice the wonderful guides eventually get them through each of the check chart activities required to advance levels, though students do genuinely do this on different schedules and may indeed be delayed for failure/noncooperation. Students can move up at each session throughout the year so this is not disastrous like being held back an entire year.
Since GT school is about to go into only its second year, we are still experimenting with the schedule and may be moving some of those activities into an optional after school hour for more depth and continuity.
Personally I hope they/we also find a way to do piano (or music broadly) more continuously, my daughters missed that one! stay tuned!
I’m planning on adding literature (and history and some other things) along these lines to the Alpha system. There’s no resistance to it, and there are patchwork solutions in place in the interim. It’s just a young project.
There are some ratios later in the piece with the same problem, and since they refer to normal public schools it looks like they must be all reversed. Is it in the spirit of the contest to let the author fix errors like that? It's pretty distracting.
They're reversed, as 4gravitons explains. There's only one place where it's stated correctly:
"At the five‑to‑one student‑to‑teacher ratio Alpha runs[...]"
The review also suffers from bizarre use of hyphenation, misuse of homophones and near-homophones ("skimming" vs. "skimping", "shoot" vs. "chute", etc.), and other MUGS problems.
The review is very informative and extremely thought-provoking, but I expected that a finalist of an essay contest would have better writing fundamentals.
> "When the data and anecdotes disagree, the anecdotes are usually right. It’s usually not that the data is being miscollected. It’s usually that you’re not measuring the right thing."
I'd like to offer an anecdote of a recent Alpha School graduate I know that supports this review (which was great btw).
I know a guy who graduated from Alpha School in ~2023. The class size was only about 10 people, so he was either a late entry, or one of the early beta testers from back when Alpha school was a new concept (thus likely not getting the full benefits of K-12 Schooling). This guy is one of the most impressive people I've met.
In his senior year, he raised ~$3 Million dollars to build a dirt bike park outside of Austin, apparently with some competitive advantage of shuttling bikers to the top of the hill and being larger than the other parks. He successfully built the park, operates it at a profit, and is now building a national mountain bike apparel brand that seems to be doing quite well. He's 19 by the way.
He told me about Alpha school shortly after graduating when he was looking for his next cofounder for his new company, and I remember I was thoroughly impressed at the time. We were both looking for cofounders at the time, so we had some in-depth discussion on past achievements, business philosophy and whatnot, meaning I got a good look into how he had become so successful literally in high school. It didn't end up working out, since we were both guys in the "driving seat" so to speak, looking for people to join our existing venture and business idea rather than looking to join someone else's.
He obviously isn't going to college, but he doesn't need to since he owns the majority of a company that returns more money than the average American makes in a year, while also owning the majority of a $2M+ piece of land. If he wanted, he could probably not work for the rest of his life, and he's 19 mind you.
Now you might be thinking; "Ok. This guy is a rich kid. His dad or family friend probably gave him millions of dollars and he built a return-producing asset with it, big whoop. It's no big deal to build a known business model with daddy's credit card." and that's the same thing I was asking him. It's impressive to build a company, but a lot less impressive to replicate a known business model when given a large amount of money to throw at the problem. Bike parks are certainly common enough, and it's not like he was building the new "Uber for dogs" so with a few million dollars it seems like no surprise he succeeded.
How he raised the money is the interesting part. He cold-called (actually called, something people don't do so often anymore) every multi-millionaire in Austin he could identify until he found one that would listen to him. He pitched the idea; secure $1M in investment, purchase a piece of land outside of Austin that's a good fit for a bike park under a mortgage, operate the bike park (and apparently there was a lack of bike parks around Austin) according to known business principles, and return a profit. He was a competitive mountain biker, and had worked at a bike park previously (I assume one of the afternoon activities that Alpha School encourages), and that was a very attractive part of the pitch to an investors.
Anyway, to sum it up he raised money, bought a property, built his bike park, and is now building his next thing. I've met a LOT of people in the startup space, like literally 500+ founders, and his story sticks out to me as one of the most practicable (he didn't try to reinvent the internet in his first company) and successful (he actually built a profitable company on his first attempt). Now you could say, as a lot of people like to do about successful startup founders, that his success was due to his unique position. That it wasn't his individual magic that led to success, just his circumstances (Bezos raised money from his parents, Elon's Dad's Emerald Mines, Bill Gates being the 0.01% of children with regular access to a computer when that was uncommon, etc.) but... isn't that kind of the point here? If that's the critique we levy against him, then that's an accolade for Alpha School! Since they're the ones who created the circumstances for him to succeed in a major way.
I don't have kids, and probably won't for the next few years at minimum, but when I do, Alpha School has been at the top of my places to send them to school for the past couple of years. My previous plan was to do elite Aristotelean-like home schooling/tutoring, possibly in conjunction with other like-minded parents, but if there's a functional model already out there, why reinvent the wheel? This review simply increased my confidence in Alpha School.
My prediction is that this review will win. SSC readers already are really interested in alternative forms of schooling, it's very well written, extremely detailed, and probably the most "review with potential real world impact" on the list of finalists.
The only bad thing I have to say about Alpha School is the branding. A-priori, the name sounds like something Andrew Tate would come up with, and its relative unspecific "We use AI to teach kids to learn!" copy on their website makes them look like the very many AI vaporware startups and/or Web3/Crypto/NFT bs. This is completely a personal issue with their branding, not substance though, and is just my personal impression, and I don't think they should change their vibe because of the association with less savory lines of business.
Edit: Apparently he is going to college. When I last had a long conversation with him it was summer after his graduation (or maybe he was intending to take a gap year? I don't remember), but apparently that changed. It still stands that he, more than most people didn't HAVE to go to college though.
I think the kid was Rhett Jones. He didn’t want to go to university but his parents insisted he go. I think he ended up choosing University of Austin (not UT Austin) as they agreed to give him some credit for running the bike park.
(I would love to read a review on the University of Austin from a ACX reviewer…)
Also worth noting that Rhett is definitely to most successful to come out of the Alpha High program. He was in the first class and he is the one the school brags about the most (they even filmed a documentary - I think this might be it — https://youtu.be/IsUwcO61irw?si=TkObs_CY34iwD2Pj )
Ahahah I wasn't aware he was their biggest success, and that definitely qualifies my anecdote. My assumption was that while he was better than average, doing great interesting things was pretty typical in the senior year at Alpha School, although I now realize that might not be the case.
I wasn't aware he ended up going to UATX, but it makes sense. It's probably unnecessary in terms of educational outcomes, but university provides structure at a time in life when structure is quite important. Wunderkind with money, prestige and free-reign do whatever he wants is a very common backstory behind washed-up adults. Having the skills and mindset to build a business is rare, and incredibly valuable, but that isn't necessarily the same skillset for a life of flourishing. Especially if you lack some other important things like humility, genuine friendships, discipline, delayed gratification, etc, which quick success early in life might make harder to develop or retain.
From my interactions he seems like a really great person though, although I don't know him personally well enough to make any judgements. Life is long, and 3 years at a unique university won't set you back if you're already well ahead of the pack.
Likes are turned off for everyone including paid subscribers (This is unique to Scott on substack). If you read on mobile you can still like comments, but that's not intentional.
Interesting. A week or so ago, I was notified that someone upvoted one of my comments on ACX. I assumed the distinction was between paid and unpaid subscribers; it's funny that it's actually because of a bug in the mobile version.
>Caplan showed that, within the normal range, nothing you do in education or parenting matters.
If I am not mistaken, didn't Caplan argue only that nothing we do in HIGHER education matters? Has he made that same claim re K-12? The decline in test scores associated with COVID remote learning (and decline in attendance) certainly implies that K-12 is more than just signaling,
That is remarkably unconvincing, since the only "data" he cites is his n=1 personal experience. And is completely inane:
> My K-12 memories include thousands of hours studying material I knew I’d never use after the final exam. The three years in Spanish were especially traumatic, but they’re only the tip of the signaling iceberg. How many years did you study poetry, art, music, history, and civics during elementary, middle, and high school? Even math often smells of signaling; honors math students usually have jobs that don’t use advanced math.
The test of the effectiveness of three years of K-12 Spanish is not whether the student ends up happening to use it; it is how proficient the student is in Spanish.
And the trope of "I will never use it, there it is valueless" is completely inane. First, although I have not used 90 percent of what I learned in school and you have not used 90 percent, IT IS NOT NECESSARILY THE SAME 90 PERCENT. A school cannot know what each student will use in the future. Second, the idea, because I do not use historical knowledge at work, the history I have learned is useless is obvious nonsense. Eg: How do know that "the Democrats are the real racists because slavery" is a dishonest claim? Because I know who the Dixiecrats were, among other things. How did I know that the claim (which I heard on NPR at the time), "the Obama Iran nuclear deal is really about oil, because the Secretary of Energy is involved in the negotiations" is stupid? Because I know enough civics to know who the Secy of Energy was. How do I know that, no, the chance of Biden winning PA and GA was not one in a quadriilliion? Because I know enough statistics to know that what assumptions underlie that claim.
And, I note Caplan, and economist does not bemoan the hours spent teaching Economics, a graduation requirement in California and, I assume, elsewhere. How odd, since almost no high school grads use econ at work. But some economics knowledge sure comes in handy in deciding who to vote for for mayor of New York.
Honestly, Caplan's take here is no obtuse that it is hard to take anything he says without a very, very large grain of salt. He certainly should not be cited in an article re K-12 education.
On Caplan's beliefs, I believe he has said basic reading, writing & arithmetic can be learned early on in school, but after that it shifts away from building human capital.
Quick note on a minor confusing thing: This post at one point says that a "mastery test" is like a final exam, but later implies that it's something that happens every day. I assume this is just a terminological error or something.
HI Taymon, I am Alpha parent who's gone to work for them. While the mastery tests are indeed like final exams, the other mastery implication probably refers to the fact that, day to day, Alpha's apps platform enforces mastery learning on each discrete skill.
Mastery learning is a core Alpha value. Most (possibly all) commercial education apps out of the box do not insist on mastery and instead pass students along with traditional "passing" scores of ~80%, which soon leaves them with big holes in their knowledge.
To be clear, I was not making a comment on Alpha School itself, I was just making an editorial suggestion for how the author of the review could make it slightly clearer.
Our family has found that incentives work, and video games are a particularly effective carrot. Our homeschooled kids have to complete their modest IXL workload and practice piano for 30 minutes to unlock 30 minutes of game time each day. They do it pretty much automatically each day with minimal complaints.
We're pretty relaxed. We figure if we can ensure our kids are literate and numerate, they'll be able to learn whatever else they ever need to. Of course, they do more than reading, writing, and math, but that's our core focus.
We used to use Kumon workbooks, but I was manually grading them so sporadically that they didn't get meaningful feedback. That's where IXL really shines--it gives them immediate feedback when they make mistakes, and this ensures mastery.
Although I am impressed by the entire IXL curriculum, we only use it for math at this point. The other major component for us is Classical Conversations, which is a community-based classical Christian program. I think the CC experience will vary considerably depending on the quality of the local community, but ours is great.
I think I would still trust data over anecdotes, regardless of what Jeff Bezos says. I come across WAY too many anecdotes from unreliable people on the internet (often anti-vaccine or COVID denialists).
I think you have to be looking at well-selected anecdotes whose provenance you understand. Anecdotes of random people on the internet are actually just really bad data.
I wouldn’t accept what Bezos says very literally, but there’s a clear phenomenon he’s talking about that is probably easiest to see in the context of “evidence-based medicine”. That particular movement says that basically only randomized controlled trials count as evidence. This meant that things like aspirin, whose efficacy was established before there were such studies, needed new studies in order to “count”.
As far as I know, there is no “evidence” in that sense that a sustained yoga practice “works” for anything, but anecdotally, a large number of people seem to find a benefit for their yoga practice. The big issue is that it’s hard to even precisely formulate, let alone measure, the benefits that people identify from their yoga practice, and then on top of that there are all kinds of statistical confounds (it’s hard to randomly assign people to a yoga practice and have it stick, and there are antecedent differences between the people who do and don’t stick with it, and there’s no way to blind the assignment to the subjects, and many of the benefits take years to emerge).
This is the sort of case, where there’s a well established practice, including serious people, where I wouldn’t accept trust the anecdote (particularly from my own experience but also those I know well) over the data.
It's true that you can't blind the subjects of a yoga study, but I think you can look at objective measurements for which the subject's awareness would be less relevant. Without that, I would still be skeptical.
Sweet Jesus, this article is almost 18 thousand words long.
Scott, let me strongly suggest that you put some sort of upper bound on the lengths of these reviews in the future. Three thousand or five thousand words should be ample for making a point or three about virtually anything. The authors can always link to more elaborate supporting material if they really want to go in depth.
Readers can always stop reading if they find the material isn't interesting enough to hold their attention. A word count limit would've eliminated Lars' excellent series on Georgism, among many other important posts.
The trouble is that if I don't finish one of the contest entries, I feel like I can't fairly vote on any of them.
Also if I don't finish a review I don't feel like I can engage in the discussion because it's inevitably "duh that thing was addressed at word 12,932".
This review started out very interesting but I lost interest in the "Does It Work?" section, whose content I could predict before I started:
You are not obligated to read every single review. That's explicit: Scott asked people to pick as randomly as possible, knowing that most people wouldn't read every one. If it's acceptable to rate reviews without having read every single one, then it's okay to stop reading a review which isn't holding your attention and penalize its rating accordingly.
"Not interesting enough to keep me reading all the way through" is a valid critique, though if you find yourself skimming or skipping frequently, you might want to get your content elsewhere.
You are correct that it's not good practice to engage in discussion before reading the entire review, but if it wasn't interesting enough to read all the way through, why do you want to waste your time discussing it?
I can relate to your reaction, but I don't think that's a popular position. Judging by their votes, ACX readers consistently prefer longer essays. The book review contest winner in 2023 was extremely long too. This year 5 of the 10 longest essays became finalists, including the three longest ones. Unless Scott sees some kind of utility in restricting the length of entries, I doubt that is going to happen.
My friend asked me to look over a text conversation he'd had, to get my read on the situation. He warned me in advance that it was "a lot" and that I didn't need to feel obligated to read the whole thing.
I said "I read Scott Alexander for FUN. Your text conversations do not scare me".
Granted, I've also read War and Peace a few times, but only because it's the perfect book for reading before bed: interesting enough that I won't not read it, but not interesting enough to keep me from falling asleep. Also, it's so long that it's practically a new book on every read-through.
Excellent piece, written ACX-style and good chances to win. I felt it could do without the complaints about their first DEI private school (but fun to read). And I wonder why you guys pays those sort of money - instead of getting a private tutor from idk Russia. Or unschool, put the school+college money in an ETF and on "graduating" your kids have enough dough for life. Numquam parum est quod satis est.
I can’t speak to author but institutional school has advantages, and a gifted school in particular has a major cohort effect. That’s part of the appeal for me, anyway.
Lol, this stuck out to me too! I was nodding along the whole time like "ah I would have loved this school so much as a kid", until I saw that and immediately switched to, "I would've flamed out and begged my parents to transfer me". I couldn't force myself to learn cursive back when there was some semi-coherent reason for it, but imagining trying to do it *now* with the full knowledge that it's entirely useless seems like it would've been even more impossible.
I was thinking at first that maybe they only need to complete a few of them, but the review seems to indicate they need to complete every single item on that list, which also seems a bit odd and against the personalized ethos.
Having seen it in practice at GT School, I believe that (a) cursive is a standin for “handwriting practice” and (b) the dosages of cursive instruction truly were not very high. None of the students seemed especially tortured by it (these were 6 to 8 year olds).
This might be a dumb question. But isn't homeschooling pretty similar to 1:1 tutoring? If the parent doing it is educated, motivated, and following something like The Lost Tools of Learning substack (or something else, I just mean a well-rounded curriculum), I would think that at least in the elementary years it would be pretty similar. Sure, it might not be exactly 1:1 (as I'd expect most homeschooling parents to have multiple children), and the parent is probably not going to be a subject-matter expert in every topic the kids ask questions about, but teaching them to research and find the answers is also valuable. You've still got rapid feedback and tailoring to the kids' needs and interests.
Why, then, aren't all homeschoolers Polgars, and why do the homeschoolers in Alpha do worse?
Hypothesis: the *typical* homeschooling parent does not put in anything like professional full-time attention to their child’s academic motivation or the wider motivational system in which academics occurs. Obviously there are going to be a lot of exceptions on the right tail, which is why you get Polgars and why some of the homeschoolers in Alpha do well, but it would explain the modal outcome.
I think a big aspect is also selection bias. I home ed because my son is autistic and I would say well more than half of the homeschoolers I know are doing so due to some form of disability; in some cases, cognitive disability.
The other issue for more religious families are large families. It's extremely hard to home school if you are also taking care of under 5s. And there's also some evidence they under-educate girls purposefully. There's also the unfortunate trend of classical education which leaves children with significant content gaps.
We also know the recent trend of 'unschooling' tends to result in pretty poor performance.
If think if you took a subset of families with a secular, academic educational philosophy, who were neurotypical, and there were no more than two kids in the family, the results would be better.
I suspect this actually is most of our homeschooling families but I haven’t dug into the data. Our primary homeschooling beta testing audience at the moment is people who say “I want to try this but cannot move to Austin”.
Points taken about the range of homeschooling contexts though.
You're definitely right in general. But also my colleague Matt Bateman is correct that the latter description fits those piloting the Alpha homeschooling program thus far.
If these smallish family & low special needs users have trouble achieving good results without the in-school motivational model offered by bricks and mortar Alpha, a general (homeschooling) audience would likely find it even harder.
Maybe parents are more susceptible to accommodating their child's natural inclination to be a bit lazier than they would optimally be. "I've done so much math today, can't I go do [other thing that is also constructive but is not math]?" is easy for a teacher to say no to, but maybe not Dad.
The academic results seem impressive, but is there any data about the effects of that amount of screen time upon developing brains, or are such concerns only made up by neo-Luddites? Or is it that screens are now ubiquitous and therefore any attempts to reduce screen time are futile? Just suck it up and accept that human brain development and eye development are potentially permanently affected? Assuming that the trade-off is that accelerated learning will mean more time for other potentially non-screen pursuits, we just might have to accept it the way we have to accept the eternal presence of microplastics in our blood and tissues.
There is no research indicating that 2 daily hours of screen time (which is not very high, and not dissimilar from typical in the majority of American public & private schools) is detrimental to elementary+ age kids.
Indications that kids under 5 should be limited to no more than one hour of screen time daily. After that age, inconclusive - some evidence pointing to higher rates of depression & anxiety in teens as daily screen time increases, but these studies (at least ones I've seen) don't typically differentiate between academic screen time and social media/gaming use. To me there's obviously a major qualitative difference between two hours of reading/science/math practice and spending your entire evening scrolling TikTok and Discord.
I think most complaints about “screen time” are really about the junk that it’s filled with, whether it’s algorithmically-curated social media or short format video, or trashy freemium games. If you could ensure that people were having live video conversations with interesting people, or watching classic movies and reading classic novels and playing deep video games, then the screen time itself doesn’t seem like it would be as big a problem (especially if there’s something making them get up and do something outside every 20-30 minutes so their eyes refocus).
Are there screen time studies that adjust for content? And speaking of breaks to let their eyes refocus(which I agree with), are there any studies that specifically look at attention span differences with regard to amount and/or content of screen time? i.e. long-form video essays vs. "shorts"?
Am I reading this right? Now, they’re collecting 40K in tuition from extremely bright children with motivated parents and teaching them in a specialized environment full of enriching externals but running at a loss. The author says the software piece is fine but basically useless without that structure. And their financial plan is to massively expand — to reach children who are not naturally bright, and do not have motivated parents, and to strip away the non-essentials, which the author says is necessary to make the program work. Doesn’t this sound like the classic tech disruption strategy, but for education? Offer a very good but financially unviable product to start, drive the competition out of business, and then offer the true, shitty, profitable product once there are no longer viable alternatives?
There are many bright children at Alpha but we don’t do an academic screen. Lots of students come in behind.
The “massive expansion” is stepwise and will occur across models with differing cost constraints and price points. I’m sure we’ll learn a lot and have to adjust a lot. The unit economics are not at all crazy. A lot of the R&D cost will *definitely* go down. The AI model costs right now outstrip almost everything else.
I'm not American so forgive me if this is a stupid question - it sounds like Alpha children complete the full curriculum in half the number of years the state expects? What do they do when they've finished the curriculum? In the UK there basically isn't any way to skip from primary to secondary school early, so if this model was expanded to the UK it seems like there would be a few years a thumb-twiddling built in. But presumably the solution to this problem is so obvious to Americans that I don't think I see a mention of it in the review.
Presumably they just start taking university-level classes? This is how it generally works in public school as well to a limited degree (e.g. AP classes).
While it is also very difficult for most bright children to 'skip ahead' in America, it *is* possible, and the main roadblocks are typically actually at the institutional rather than state level (because most institutions do class-based instruction grouped by age & ability, they do not want kids getting more than a year ahead because then the kids will be with much older children who are not at the same level of social and emotional development).
After that the hard part is basically just finding a university willing to accept a young student (though this does happen, so clearly it's not impossible) or alternatively building a post-curriculum after the state curriculum is completed.
(I am parent of Alpha students who's gone to work for them) Not stupid! - basically high schoolers begin to move in different directions. Some students are set on entering very competitive colleges, for them doing the minimum to have finished high school equivalent is not enough, they continue to do math/literacy to prepare for SAT (college entrance exam) and Advanced Placement coursework (college equivalency program). Others move more in an entrepreneurial direction, for instance.
The gifted and talented version of the school essentially aims to have students college ready academically by the end of 8th grade though it's new enough that I don't exactly know what the first classes of graduates (several years away) will in fact choose to do. Some probably will enter college before age 18.
You can actually go to college a couple years early in the US without any significant difficulty, and earlier than that with some more effort and arrangement with the school.
For anonymity reasons you should probably blur out the name in certain screenshots; anyone familiar with a certain educational Substack would recognize it.
I didn't read the whole thing, because it is way too long and not particularly interesting to me personally. I stopped about halfway through and skimmed the rest. However, from what I gathered, this seems like a great program for gifted kids, and not so great for other types of kids, especially ones that would struggle with the pace. But then again, gifted kids will do great in just about any type of program, especially if they are self-motivated. So what's the point in spending $40,000?
I think the point is Alpha is self-paced so they go faster with less time wasted on bureaucratic busywork. It leverages AI in a way that semi-simulates a 1-on-1 tutor experience. It's an improvement over traditional private schools which people have already demonstrated a willingness to pay 40k a year for.
It seems promising but my guess is the real secret sauce is selection effects and enthusiastic founders. Wealthy early-adopters are a highly-selected cohort, as are educators willing to take a chance on a radical new teaching model. That's unlikely to do well at public-school scale, particularly in misbehavior-prone urban districts. But it might be a viable alternative to both magnet schools and traditional private schools.
The quote is deeply ironic in that the point of teaching the child is for the child to learn, which using AI undermines. The point of the teacher teaching is for the child to learn, which using AI for that half does not undermine.
I thought the 5:1 teacher to student ratio was pretty crazy, then he mentioned public schools having a 25:1 ratio and I realized what went wrong. I could see some school for rich kids that has five people teaching each student, but I've certainly never been to a school with 25 times as many teachers as students.
As homeschoolers we do some of the same things; we use anki (which is free!) for spaced repetition, we do a sustainable amount of academic work daily rather than pausing for the summer, and of course we don't have to try to educate thirty kids at a time. These make learning vastly more efficient and Alpha School didn't invent them. But we also do everything we can to surround our kids with adults who are enthusiastic and open about their own lifelong learning (ourselves included) and with materials that are worth their time (a thing I wouldn't say about IXL).
Humans are very driven to learn, care what the people around us are doing, and are also mostly at least a bit demand avoidant, so it has always seemed to me that homeschooling parents can choose between spending their time explaining why kids have to do math but adults don't, doing some tricky alcumus problems themselves, or paying their kids and hoping the kids don't find a better revenue stream.
I have no issue with paying kids to do things that they are free to refuse to do, but as a kid I would have found it insulting to be required to do something and then paid just a bit for doing it. I understand that other kids are more pragmatic though, and are happy with the arrangement. I also would hesitate to pay a kid to read a book because there is no clear definition of reading a book well enough, and I'd rather read some of the same books they do and enjoy talking about them together than quiz them and incentivise the opposite of close reading while becoming the did-you-actually-read-it-though inspector.
This is doubtless racist, sexist and homophobic, but Alpha seems to have a lot in common with schooling methods during lockdown. And Bastet knows that those methods were disastrous for educational outcomes.
The one exception I know of personally was one ant colony, and that was simply because the queen ant took homeschooling to insane levels. Before, the little ants had been begging to be homeschooled. Soon, they were begging to be able to go back to school.
Which brings me to my real point - for decades now, humans have been pushing various nostrums to address the fact of Why Can't Johnny Read? None of these nostrums and magic spells perform as advertised, although many of them do enable educational consultants to make bank.
Fact is, parental involvement, parents that ride little Johnny like they are trying to win the Preakness, combined with an environment that stresses learning, is the only thing that works. Nobody wants to hear that, because other than that queen ant, nobody wants to put in the work. Nobody wants to point the finger at the parents. Far easier to look for magic bullets.
There are some really obvious cases of ChatGPTese sprinkled throughout this review, yet it's also clearly thoughtful and original, with lots of concrete and specific claims. Very unusual combination in my experience.
I hope once the contest is over and things can be de-anonymized, the author would be willing to comment on how they use AI in the writing process yet produce a high-quality non-slop document. Although I do worry the answer is just "you have to be literate, and also actually care".
ChatGPT has a writing assistant mode. You can tell it what your central idea is and have it produce a rough draft. This guy probably did that and then rewrote it, leaving some of the original in.
"Elite private school attendance buys you smaller classes, brighter kids, and fancier field trips – not academic acceleration ... That prior—show me velocity, not polish—is the lens through which the rest of this review should be read."
The highly telegraphic writing style and punchy parenthetical phrases in italics (“show me velocity, not polish”) are in my experience extremely characteristic of o3.
Excellent article. I'm a bit more skeptical than you are on the magic of Alpha School compared to other schools, academically, and its ability to scale to a true mass market product. However, the tech-heavy model does seem like a way to cut expenses ($40k flat) compared to Horace Mann ($64,070 + a lot of extras + an endowment) or Harvard Westlake ($49,700 + a lot of extras + an endowment) for an elite school, and even if they have to drop some of the international travel for grade schoolers, it might be a sustainable long-term business without billionaire financial support. I hope it's still available when I have kindergarteners.
$40k/yr for 17 years to have a 95th percentile achieving kid is an incredibly good investment.
Quick story about another student who goes by Austin Scholar here on Substack.
When I first met her, she was a sophomore in high school (15 years old). I was teaching writing at the time and actually rejected her from my program because it didn’t seem right for a teenager to participate in a program designed for high-powered adults. To her credit, she pushed me to get in the program and after a short interview, I reluctantly let her in.
There were ~250 students in the program and she was a top 10th percentile writer as a high schooler. Watched her get a perfect 1600 on the SAT and get admitted into Stanford. Even as I’m writing this, she’s doing an internship with me for the summer and one reason why I’ve been eager to hire Alpha students for various projects is that they’re uniquely self-reliant because of the lack of teachers in school while also being uniquely on top of things because of the strict standards that Alpha maintains.
All this is to say that Alpha doesn’t work for every students (and it’s not designed with that in mind), but when it works, it works pretty darn well.
As someone who has done casual inference professionally for a very long time, I can say with confidence that Caplan's work is an exemplar of Gelman's many-forked path, where the results were mostly determined beforehand and confirmed by choosing the right path through the data. Not work to be taken very seriously.
Basically the things he's trying to claim are very hard to identify in the statistical sense. And extrapolating even from studies where ~everyone would agree the causal inference is solid is always low confidence because of shifting context and the specifics of the outcome. This means that it's a matter of judgement for every analysis about what the results are actually telling you. This would make the kind of sweeping statements he makes suspect no matter what. (The deep dive into a specific analysis of his is what I don't have time for).
But there are a number of other clues that you shouldn't believe his judgement calls aren't always strongly biased. First, his shocking lack of technical skill. His anti education book was based on 15 months of building an opaque tangle of spreadsheets. Which he then complained no one double checked and reanalyzed 🤦♂️ This alone is near disqualifying. It's almost certain he's doing mediocre-at-best casual inference when he's limited to what's available in excel. It also makes his work opaque and pragmatically unreproduceable.
Second, he writes a mixture of opinion/ideology, data reporting, and results from his own spreadsheets. He will often try to pass off his ideological commitments as facts. Essentially I'd argue that he's revealed his commitments. Everything he does is work to bolster his ideological commitments through data mining and argumentation. He spends a lot of time on polemic and constructing logic chains that he thinks make his analysis solid. But he's never spent the time gaining the technical skills to actually do best-in-class non-experimental casual analysis. In the same way you should be skeptical of left wing academics whose work always just so happens to "prove" that whatever flavor of leftist campus activism is hot right now is true, this shouldn't give you a lot of confidence in all the methodological choices he makes in constructing his data argument.
Alpha's marketing is excellent - and there are many educators and parents I admire who are excited about them. I've been craving more detail and this answered many of my questions.
On the pedagogy of app-based learning & measuring using MAP:
This approach seems solid for the basics of math and reading, though this piece's point that reading and discussing longer form books is missing is key. Much has been written about the negative impacts of over-optimizing for short passage reading on the literacy skills of American young people. The short-passage only approach is usually associated with public and charter schools, while the Harvard Westlakes of the world go all-in on literature.
Science is even trickier to teach via apps and measure via standardized tests. Many schools don't take MAP Science at all, as a note. IXL works well for learning baseline science content, but I have a hard time imagining multiple-choice modules go far enough to produce mastery of science skills as written in the Next Generation Science Standards. You can read those on their website, but they are largely complex and applied skills such as 'developing models,' 'representing and graphing data,' and 'making evidence-driven arguments'. IXL teaches these skills by having learners 'fill in the blank' of a partially completed graph or argument. The MAP Science test does the same, because it's also a multiple-choice measure. Isn't this fill-in-the-blank mode of thinking much easier than asking a learner to actually demonstrate the skill (figure out how to represent and graph data) themselves? If I were an Alpha parent (especially one not at the GT school with the extra academic competitions, which sound great), I would worry the science learning isn't rich & applied enough.
Not sure how History/Social Studies learning is assessed at Alpha, since MAP doesn't have a Social Studies test. I assume just in-app and their mastery tests. Similarly to science, I would worry IXL + a sprinkling of reading passages in AlphaReads lacks the richness and depth I'd want. To learn Social Studies well, I'd want them to follow a more comprehensive trajectory that includes longer sources, group discussions, research projects, etc. I'd worry that the afternoon activities as described are too piecemeal to succeed in this area but could be wrong.
Not surprised Writing is a miss in this parent's eyes - I've never been impressed with any edtech writing app. There are a few that are nice for drilling grammar and can supplement writer's workshop style teacher-led classes.
These things feel surmountable. I wonder if Alpha has considered using a portion of afternoon time for teacher-led, Hoel-style 'aristocratic tutoring' in humanities and/or science to see how well these in-app learning gains transfer to real-world applications and open-ended conversation, while remaining hyper-personalized.
Also, I was surprised to see IXL as the app of choice. The questions in IXL are quality, but it's boring as hell - the forms of engagement are unvaried and dry, and it must feel repetitive for kids to learn in the same mode every day for up to 12 years. Last time I spoke to somebody at Alpha, they said they used several apps for each subject for variety and to ensure coverage - perhaps this has changed, or maybe just wasn't mentioned here. They'll need to find a way to make the incentive system foolproof for parents to implement in the homeschool/virtual model.
Early in building Prisma (virtual program I work for), we experimented with an app-driven morning schedule. We ended up retiring that approach for every subject except math, for some of the reasons above. This year, our kids grew 2.3x as fast as their peers nationwide in math, and 1.6x as fast in reading on MAP. We definitely could read this as evidence that apps are the best method and should be applied to the other subjects. But our parents tend to choose us for reasons other than optimizing these scores, and would hate if we abandoned our discussion and project-based Literacy curriculum for IXL. Could we supplement with apps and boost those further? Probably! Overall I'm excited Alpha exists. They're developing some cool innovations that can meaningfully improve the space.
This article was a great read, and I agree as a parent of GT School (Alpha’s school for gifted children in Georgetown). My son is 10 and taking 6th and 7th grade classes. He met and/or exceeded all of his end of year MAP goals and continues to be in the 96-99th percentile across the board (one reason we entered GT School to begin with). He also has developed a true love of school. To the point where he sometimes complains of we take him out for a day for a long weekend. He especially loves his guides, who routinely find ways to customize rewards and incentives to each child’s interests. My son won a Lego chess set and Mario stickers amongst other things with his GT bucks. Their reward system is highly effective and the personalization acts to make students feel seen and understood. Though the author is correct that I don’t think this model will work for every child, my son has ADHD and is actually doing spectacularly well even without his pervious IEP accommodations. Their low ratio and individualized learning model have been both impressive and effective. This all coming from someone who was highly uncertain and critical of the idea of AI teaching my child.
The website has a much weaker 100% money back guarantee than the one stated here. Can anyone find a link to it? Perhaps they've walked it back as guaranteeing every child wins a national award seems like a good way to lose your shirt.
I see people are interested in hearing from the parents — I am one, and I'm NOT employed by the company. My 7-year-old is at the GT School (which, btw, is $15K/year; Alpha's specialized academies have lower price points). AMA.
To start, here is a bullet-point list of why we drive 30 minutes to the school (TL;DR: it's worth it. We will move closer now that we know it works for our son):
- Compressed academics utilizing the Pomodoro technique — work intervals followed by breaks; much more effective for high-energy kiddos who can't yet focus for extended periods of time
- Self-pacing — go as fast as you can or as slow as you need in each subject; no shaming through grades, no "B student" stigma
- Verification and accountability — tests are used in a productive way. Apart from three-times-a-year MAP testing, there are mastery tests whenever the student completes the lesson plan for each grade. I wasn't a big fan of the testing part going in, but they use the detailed data from the test report to identify gaps and then serve additional supporting lessons in those areas. My son completed 2.5 grade levels in one year, but I'm more wowed by the fact that he actually mastered that material than the speed itself.
- The guides!!! — well-paid, highly motivated individuals who are not burned out from a mountain of paperwork and can truly engage with the kids
- The group workshops and individual pursuits (check chart) — the element of real-life skills, learning how to bring a project from concept to fruition, on your own or with a team — this part was really important to us
- Growth mindset environment — "the power of yet," philosophy of high expectations with high support, finding your "why"
- Community — engaged parents who are willing to take reasonable risks to try something different for their kids
It's not the AI, it's not a magic bullet — it's not one thing; it's the whole package.
And before you say, "Your kid is already gifted, so he would do well anywhere," that is a common misconception. Gifted does not equal high-achieving. Our boy is extremely high energy, motivated mostly by his interests, with spotty attention-paying ability. We are quite certain he would not thrive in a conventional schooling setting, and we have spent countless hours researching and considering various education models. This is not perfect; it's just the best we found.
Also, just for fun — and also to address the "wealthy kids" stereotype — we fall into the "first-generation immigrant family" pool, making sacrifices to give our son the best shot. I know every parent wants that for their kids, and I'm hoping my long comment will be somehow helpful in your journey.
I am the parent of two Alpha students, they are moving to the GT school this fall. As for me, I've been drawn out of more than a few years of stay-at-home motherhood in order to get involved in the Alpha mission.
We initially chose Alpha just because all the other schools available schools seemed blah. My husband, Byrne Hobart (an independent finance and tech writer) is one of the non-conformist types mentioned in the piece.
But my older daughter turned out to be exceptionally bright and her results at Alpha have been way beyond expectations. With no homework at all and very little drama, she has maintained not only 99% achievement but also top several percentiles of growth for 3 entire school years.
I never heard her complain that she was bored, never had to meet with a teacher to grovel or battle for acceleration. At home we focus on reading for pleasure, arts and crafts, and exploring niche interests (she's obsessed with penguins).
Intense acceleration offered in a pleasant environment has been a priceless gift to our family. I never worried about oldest's education while I was pregnant again and caring for an infant.
If it means anything to you, this is what o3 says when offered my daughter's MAP scores: "[X] is operating at an 11th-grade academic level while still in Grade 3 and is continuing to out-grow almost all comparison groups. Alpha.school’s 2-Hour Learning model is not only keeping pace with, but in several dimensions outperforming, America’s most elite K-12 institutions—all while freeing up 60-70 % of the day for deeper projects and community-building, exactly in line with the school’s founding mission."
My second daughter joined Alpha a year ago and she's a tougher cookie (2e/on the spectrum). Even with far less than 2 hours per day on the apps (despite her wonderful guide's ceaseless efforts and encouragement), her growth has still been immense: https://x.com/gtmom/status/1931058079745056974
Very happy to answer any other questions, here or privately.
meh, nothing about this is particularly different from other educational fads, because they all share the same commonalities:
1. cherry pick the most gifted and motivated students
2. target parents who are both rich and overinvested in their kids education.
the thing is, those two things can make any kind of system work. I see this with homeschooling; everyone glazes it the fuck up, but its horrific otherwise if the kid or parent just can't throw heroic levels of energy at it. That's when teachers help, not systems per se, because its very easy to game a system. Hence chatgpt writing papers and passing around test answers. or cheesing curriculum requirements.
and honestly you all overrate intelligence and underrate connections and money in life success, that is why your kid does well or not short of mental illness or parental tragedies. Education is probably not going to help you beyond a point, and there is some truth to Caplan.
(only at the extreme levels, though: its annoying he is so stupid and doesn't realize he needs a widely educated base of people to keep his university running, and that tenure hothouse people give far less value to life than he things)
#2 is fairer than #1. Alpha admits plenty of students who are behind academically and not especially excited by school. GT School screens for academic potential, Alpha absolutely does not.
Re: cheesing curriculum requirements, the students sometimes do try to do this, and it gets patched. It’s a fun arms race.
I don’t think the premise that “any system works with rich kids” is particularly true. A lot of very specific works goes into helping these students, and I don’t really see evidence that it’s just a wash compared to very different methods at other schools, and see plenty of evidence that the school is helping these students. (Other elite schools don’t even publish their K8 testing data.)
behind compared to who, though? parents who give kindergarteners IQ tests? other elites? i really don't think there is that strong a difference, and that's not getting into parent differences in how tiger mom they are.
the cheesing i just mean that without motivation by some one systems fail. id bet the weak point is the brazil teachers, lol at tech bros offshoring. better hope they don't fake answer calls, or exploit the system because they can.
and no those kids would be fine without that school. places like alpha bank on that, because its really hard to actually teach struggling students. little effort of maximum profit is the good way for modern business men
There are multiple comparisons of interest. I think with Alpha in particular the most helpful comparison is other elite schools, yes.
The “brazilian teachers” are software developers and academic specialists who work on the curriculum and platform. They run the coaching calls because they are the ones who need to know what’s not working and fix it on the app side. Not sure about Brazil but some of them are indeed remote.
> meh, nothing about this is particularly different from other educational fads, because they all share the same commonalities:
This seems like an oddly defeatist attitude. Everything you try will appear to work, therefore there's no point in trying to figure out how to do it better.
> and honestly you all overrate intelligence and underrate connections and money in life success
This I agree with, but it does seem like Alpha School is aware of this and trying to deal with it, with "life skills" (of the sort practiced by mid-level managers at large tech companies) being a large part of the curriculum.
they arent trying to do better, they are trying to do cheaper and get quick bucks.
like you don't need to keep reinventing a stable solution; a private school with correct teacher/student ratios and managed well is perfectly fine. But to run one is expensive and modern businesses love their mvp and getting something for nothing. They'd love it if your kids teach themselves with ai lapdogs and the occasional overseas (got to pay as little as possible!) check in.
honestly businesses are not looking out for your own good.
the life skills...eh its a skill to know the right people i guess, my point is more the kid being smart or not matters less than the dad being famous, which zip code you live in, if dad overloads you with extracurriculars, or if parents can bankroll first business or internship. or failure might be from things education cant help much with.
think the chinese dudes know with guanxi, but we kind of love our horatio alger.
> like you don't need to keep reinventing a stable solution; a private school with correct teacher/student ratios and managed well is perfectly fine
It's perfectly fine, but is it optimal? My K-12 education was perfectly fine, but could have been better.
> the life skills...eh its a skill to know the right people i guess, my point is more the kid being smart or not matters less than the dad being famous, which zip code you live in, if dad overloads you with extracurriculars, or if parents can bankroll first business or internship.
Right, that's exactly the sort of life skill I wish I'd learned at school and didn't. I emerged from school with too much faith in the idea that just "being smart" is the route to success, and ignoring all the other aspects.
I, too, felt some instinctive revulsion to the idea of incentivizing children to learn, and I think I know why it feels wrong: it might actually work.
No matter how you dress it up, school must involve adults compelling children to behave in certain ways. People who like freedom find this somewhat unpleasant to think about, but most agree it's for the greater good. But incentives are just a much more effective means of compulsion than other methods. At least when a teacher hits a student it's clear what's going on. Incentives are, well, seductive.
Reading this essay makes the whole school feel very creepy. Points for compliance. Eye-tracking software. When we're worried about socialization, we set up a new system to reward that too. The kids in this system don't really seem to have free will -- they are allowed to make choices, but only within the system.
I don't think this is necessarily a bad thing! If we're going to control how children think and behave to educate them, we might as well do it effectively. This school sounds a lot more fun to be in than the one I grew up in. But it is very much the Brave New World to the public schools' Nineteen Eighty-Four, and it's hardly a surprise that a lot of people find that scary.
Fascinating read. Really appreciate the context outside the school as well as the deep dive into the entire methodology.
There is one big, conformity forced assumption here and that's competition. Absolutely everything in western society is a competition and we've been taught that is the best way and the only way to live. Our kids are measured by percentiles and only if you are the best of the best can you ever hope of being accepted into the best colleges and once you have the name of that institution on your resume then you can work for the best firm to make the most money and then you'll be happy. The problem of education is always framed as "how can I make sure my child WINS?"
For all the talk about nature vs nurture and Caplan's work, this entire post reads like signaling. “Insofar as schooling makes you healthier by raising your status, its health benefits are zero-sum: you can’t raise your rank without dragging others down” (Case Against Education, p.171). I don't believe your claim that after love and respect the best thing you can give your kids is time. Or rather I believe the claim, but I don't think you believe it. Why move your family across the country in the middle of the school year and spend $120k a year on that schooling? You didn't do it to give your kids time. You did it so they can make the same kind of money.
How do I know it isn't for time? They don't even have a free half-day. Chess competitions, quiz competitions, essay competitions, property management (I'm so curious to see how a bunch of 5th graders would decorate an Airbnb). Even if you went to the sports academy it's the same game. Parents will spend hundreds of thousands of dollars on sports training to beat other kids. Beat all the other kids and they can go to those same colleges to get those same jobs. A huge percentage of Olympians in the 2024 games from around the globe were currently attending top American Universities. In that context as you mentioned over-optimization runs the show. And if they finish school before college age are you telling me they will actually spend 7 years of their life with free time to do what they love? Maybe some will, but the kind of parents enrolling in this school are likely going to push their kids to finish college early and make money earlier.
This kind of schooling may require a certain kind of non-conformist to enroll, but you are still conforming. I'm certainly not telling you how to parent and I'm not saying money doesn't matter. I'm saying is I don't experience any happiness from "keeping up with the Joneses".
At GT School, competition is opt-in. The students find it fun. It’s a culture of *friendly* competition. Really remarkable—that culture is actually the thing that sold me on the school.
I completely agree. My son enjoys the competitive aspect, but also excitedly tells me about his friends beating him. GT School is comprised of high achieving kids that like competition but the guides are very gifted at making everything friendly and cooperative at the same time.
One of the main reasons we're sending our kid to this school is because this model is not based on a zero-sum game like - I would argue- most conventional schools are. I can't comment on the author's ambitions for his kids but ours are simply that he finds an occupation that brings him joy and fulfillment and finds a way to support himself and his family doing so. Many ways to skin the cat, many ways to live a happy life and this model offers more time and freedom to explore those ways, while still maintaining accountability for the basic core skills, than others that we looked at.
Fair enough. I'd probably send my kids to this school too simply because I'd want them to learn as much about the world as they can. It seems like they really excel at that. And no matter my objections it's not like other options are going to be so drastically different for the end goal. We're all just doing the best we can with what we can.
My frustration is more about the problem we are trying to solve here. The post is focused on academic measurements, percentiles, and the marketing material feels aimed at winning the zero-sum game. Learning itself isn't. But as the author points out the natural next questions are about getting into a good college and getting a good job. The questions almost all parents inevitably ask themselves.
I recently talked to my cousin who went through an accelerated program, started college 2 years ahead. I asked what she loves about her chosen field/major, basically what drives her in school. She said she doesn't like it, but it'll make good money. Her parents and every single adult in her life talks the same way. It's always about making as much money as possible as early as possible.
It may be impossible to not play the game no matter how pure your intensions. It's just so culturally ingrained. Personally I'd want us to solve the problem of "loving to learn". In many respects I imagine this school does a great job of that. However I often think of Goodhart's Law: When a measure becomes a target, it ceases to be a good measure. We are measuring academic performance relative to other kids. Our hyper fixation on comparison in academics or on social media can really suck the joy out of living if you let it.
I loved this essay but your comment here captures most where I land.
We humans with fairly finite lives can only serve one or two priorities at a time. If getting ahead, faster, so that one can "win" the money game is the priority, it will crowd out quite a few other good potential priorities.
In the video of the little kids each talking to their own individual AI about how to do their hobbies, one of the boys asked the AI, "And what do I do now?" The AI was instructing him step-by-step what to do to for his arts and crafts. *To what end?* What happens when we can't really guide ourselves any longer?
I really liked this essay and would also love to send my kids here if it was cheaper. But I agree there is something deeply unsettling.
I also read thezvi’s childhood roundups, and while I like those they make me bristle at times about the absolutely relentless min-maxing. I have been considering posting a comment similar to yours on Zvi blog actually.
I feel like the optimization bug has infected me and sucked the joy out of my life. I suspect it became ingrained over a million hands of online poker, or maybe it’s an OCD thing. I have to justify everything in terms of EV and I hate it. I don’t want to pass this on to my kid.
I was caught off guard when the author mentioned using Duolingo to teach foreign languages. I thought it was fairly well established that Duolingo isn’t particularly effective for learning new languages, due to the lack of active recall, among other factors.
Yes. Take that insight about Duolingo, and consider that the other educational apps mentioned are similarly engineered. I wrote a lengthy comment elsewhere in this thread elaborating, but this is a real concern - does bite-sized, multiple-choice module learning transfer meaningfully to an authentic context as well as it does a similarly designed standardized test?
There's evidence that it may - SAT scores are the best predictors of college achievement, for example. But I wonder if Alpha students' test scores represent their real-world intellectual ability as accurately as a kid who spends much less time on 'test prep' style learning.
Maybe Alpha's mastery exams are designed in such a way to mitigate this, too, I don't know.
We agree that Duolingo doesn’t work. The students wanted to try it last year at GT School for various reasons, but it’s not part of the platform.
Re: transfer, overfitting to tests, and achieving deep understanding, I do think that’s a risk area for us. We have a different pedagogical risk profile than most other alternative schools. I think we’re managing it well but it’s ongoing. I definitely don’t think it *trades off against* the methodologies we use, but we have to be sure we’re achieving it.
It was made particularly worse by the fact he complained about the effectiveness of language classes in regular high schools with human teachers. And he brings up Duolingo as a better alternative????
I agree. Duolingo is useful for vocabulary memorisation, (very) basic reading comprehension, and habituating the learner to a range of native-speaker accents. But its commitment to bite-sized chunks of content prevents it from teaching grammar in the clear and explicit way that's necessary for language learning. There are many grammatical concepts that simply cannot be taught from scratch in five minutes flat.
This is one of the best things I have read all year. But I can’t help being just a bit upset.
As someone not involved in education, it has always seemed to me that an approach similar to this is, as kids used to say, “a no brainer”. If I had a billion dollars and wanted to design a school to optimize learning, it would look something like this. This is what I remember worked best when I was in school, and new technology would just empower it more each year.
The reason I am a bit upset is because I have heard for decades how “nothing works or scales in education”.
I am starting to think that the problem in education isn’t the children, it is the dysfunctional mindsets and frameworks of most people involved currently in education.
Obviously this won’t work for every kid, and is currently not affordable for most, but I think something like this could soon make a difference for millions of families.
"Some kids will rush into college classes. Some will choose to use the time to play sports. Some will use the time to master chess or quiz bowl or programming. Some will take time to travel the world with their families."
If we go by the South Korean case study, ~all kids (or all kids of ambitious parents) will choose to spend that extra time on more and more classes, as competition for "elite" colleges and the "elite" jobs they entail only gets tougher and tougher. In the end this just increasing the price of signalling even further, leading up to a very expensive equilibrium.
Not all kids! My son enjoys the 2 hour learning model because he is bright and that’s all he needs, but also because he finds school work boring compared to workshops. He aspires to be a gaming programmer and we won’t pressure him to attend a fancy school to do so. Wealth doesn’t always equate to elitism. I’ve found that most private school parents just want happy, thriving kids, just like any other parent.
That sounds wonderful and I don't doubt for a second that this is a great environment for a kid right now. In fact, if I was living in Austin and had that kind of money, I'd also consider this school for my son - mostly for the wide plethora of cool and exciting things the kids seem to be doing in the afternoon according to the review.
What I am skeptical about is how such a system will evolve at larger scales (if it is as effective as advertised). Consider this scenario:
You start with Alpha school with 2h learning in the morning and workshops in the afternoon. Then someone comes up with Beta school that uses the same method but ramps up the learning to 3 hours to get 40% higher test scores (let's assume some diminishing returns). Gamma school is at 4 hours, Delta at 5. The alumni of each of them crowds out the alumni of the previous one from elite institutions, providing constant selection pressure. By the time you know you have rooms full of kids glued to computer monitors for 8 hours a day doing spaced repetition pomodoros.
There is currently no market for this. Most parents, even parents shelling out for Alpha, believe that the 2x learning promised by 2HL is already possibly too much/too fast, not too little. The scattered parents who are like this satisfy their preferences with tutoring, math camps, etc added onto other private or public school educations. They do not, in the U.S. at least, form an interest group that drives whole schools into or out of the market.
Though I do wish that there were some secret cadre of academic accelerationist families ready to book out GT school and pressure us for even more right away. Would be a great problem to have, to be banging the drum that their students need life skills and not just academics, rather than explaining how little students learn at regular school over and over.
Phenomenal piece. It highlights a point that any rational observer of the lock-step "product" of American education might ask: why not let kids progress at their own pace, tailored to their ability, and focus on mastery instead of simply promoting them annually by default? You're not claiming to have a gold-standard study here, but honestly, it’s not hard to imagine how well that kind of model could work.
That said... I felt a real sadness reading it. The image of kids drilling on computers—even with breaks—was disheartening. My reasons are entirely anecdotal, and I can’t back them with empirical evidence, but I went to a remarkable private school from preschool through 8th grade. My cohort in the "Honors" track was made up of extraordinarily bright, talented kids who later became academic, economic, and artistic achievers. Total selection bias, no doubt—but here's the thing: we loved school.
Why? Because our teachers loved teaching. History came alive for us every year—from the Egyptians in 4th grade to post-WWII in 8th. My lifelong love of literature was sparked by our English teachers reading aloud with passion. If you've seen Dead Poets Society, it felt like that. We wrote and wrote. We adored our science teacher most of all—my closest lifelong friend, now an M.D. Ph.D. and Chief Science Officer at a biotech firm, credits that teacher with changing his life. Another close friend, a composer, points to our Head of Middle School—a master musician who packed our days with Glee Club and Orchestra—as his inspiration.
I don’t quite see that at Alpha School. There’s no shortage of engaging extracurriculars, and maybe kids just self-select into what resonates with them. Still, I don’t see love and passion. And I get it—love and passion don’t scale. They didn’t even scale within our own school. The honors track received all that energy; the median cohort—still smart, still tuition-paying—got "good, not great."
Maybe what we had was the modern equivalent of a Victorian-era tutor: personalized attention in a small enough class where everyone could thrive.
The love and passion are there. I wouldn’t be sending my daughter there next year if it wasn’t. It’s hard to capture and convey in the same breath as one is talking about the software platform, but they are there.
I get the worry about wanting inspiring teachers and role models, though. I had this worry too, before I visited. The teachers do serve this role, it just has a different specific look than it does in Dead Poets Society.
OK, that is wonderful. It would be interesting to try and quantify the "inspiration" effect. We talk about how the data says, well, it's mostly your genes, and maybe your peers. But so there is so much anecdotally about inspirational teachers and professors, that maybe "the data isn't measuring the right thing." (OK, maybe in this case not measuring "all the right things." Appreciate your response.
We have a survey question we ask to students annually: “Every adult had 1 or 2 teachers who changed their life. Is your teacher that for you?“ A crude measure, for sure. But that’s the aspiration, and we take the answers seriously.
> If all of this makes your inner Bayesian flinch, you’re in good company. After twenty‑odd years of watching shiny education fixes wobble and crash—KIPP, AltSchool, Summit Learning, One-laptop-per-child, No child left behind, MOOCs, Khan‑for‑Everything—you should be skeptical.
Wait. Can someone please explain to me how KIPP failed?? AFAIK it's done great for low income minority kids.
KIPP lost its "no excuses" orientation and went a bit "woke" like many other schools but as far as I can tell the academic results are still notably good.
I am not familiar with the software used for other subjects, but Duolingo for foreign language instruction for $40k a year is very funny :) I hope you do not expect your children to actually learn any foreign language as you would be disappointed.
I thought that was interesting too. I've used duolingo quite a lot after having learned a few languages the old way previously. It's a really impoverished platform. I started using Pimsleur to compare and it's a much much better program, if you're doing language learning without a live person. But language is a human communication tool -- it seems very odd to me to not learn it through human communication with all the nonverbal cues and emotional involvement that come along with that. To properly learn another language is to absorb another culture and history into your bones, and to some extent to become a somewhat different person in the world.
It feels to me like there are a bunch of deeper layers of learning that happen simultaneously when solving a math problem live with a group of people or discussing history or talking about ways to interpret a piece of writing than can be taught through an algorithm and tested for with multiple choice quizzes. The affordable, scalable two-hour part of this seems much like duolingo to me, which is to say impoverished. And the afternoon enrichment looks much like any other kind of enrichment available to rich private schools.
The good news is that we don’t use Duolingo for foreign language instruction and we agree it doesn’t work. It’s something the students at GT School wanted to try this year.
The bad news is that we don’t have a solution for foreign language instruction at present.
>Workshops in the afternoons are the “fun” part of school. They are the equivalent of the music, theater and art classes that fill in a traditional school schedule (just more focused, measurable and creative).
Am I hopelessly crunchy-granola for thinking that if Alpha is willing to dedicate a whole afternoon program to sports, it would make sense for them to have comparable humanities program? Maybe it’s not “measurable” enough, but engaging with art at a young age seems at least as valuable to me as knowing how to operate an AirBnB.
I agree. And when it comes to the literature part of the humanities, I don't see how computers can be much help. The thing about the classics is that many of them are hard to access -- hard to find your way into and like. If you can't find your way in -- if Dickens just seems slow and boring, if Shakespeare just seems like a jumble of way way too many unfamiliar words -- then you need someone who loves those writers to show you the way.
There's really no point in forcing yourself to read one of these things if it bores and vexes you from beginning to end, just so you can say you read it. These are pieces of literature that can give immense pleasure. If you can't get over the hump and don't feel any of that, then even if you read it and can answer College Bowl questions about it you haven't read it.
I homeschooled my daughter through 6th grade. Based on my experience, I wanted to give people a data point when it comes to acceleration and speed of learning.
I was extremely laid back regarding book learning until one year before she was going to return to school. By age 5 or she taught herself, with a little help from me, to read and write. (But her spelling was intuitive and phonetic, and her handwriting was very messy.) She picked up a modest amount of add-and-subtract math and simple geometry from games and Montessori type toys. But that was it when it came to the 3 R’s. She did read a lot during her homeschooled years, and was involved in various activities that enriched her mind and life but probably had little effect on testable academic skills.
During the year leading up to her returning to school we worked 2 or 3 hours almost every day on the 3R’s: Spelling, grammar and punctuation, math, handwriting, and expressing ideas in paragraph length bits of writing. I used conventional textbooks and workbooks popular with homeschoolers. Neither of us enjoyed these sessions much, but they worked fine. By the end of the year she was up to speed on all these subjects. I did not attempt to teach her school curriculum science, history, geography etc., since it seemed to me like mostly she would be learning new stuff in those areas, and would not have trouble filling in any gaps in her knowledge if needed.
So she and I sort of did it backward. Instead of getting her way ahead academically, so that she could finish her formal education early and then be free to begin a career or do other things, I let her be free to do other things for 6 years, then taught her the essential skills covered in school in a tiny fraction of the time that would have taken in school. Of course she learned lots of other things during her homeschooled years, and if I wanted to explain more about why I homeschooled the way I did I would describes these. But it’s not relevant to my main point: Teaching the 3 R’s does not take anything like the amount of time schools take with the task. And if you wait til kids are older it far easier for them to learn it — in fact it is extremely easy for them. And my daughter is not one of those intellectually super-gifted kids. In fact when she was small she had some kind of low-grade spoken language disability. (I adopted her from China when she was 2 years old, and language delays and problems are very common in foreign-adopted kids.).
I understand that my entire take on child education is different from the one being described in this review, and am not here to argue about that. Just wanted people to have this data point about how rapidly a kid can master basics, especially when older. You should also know that even the homeschoolers I knew who were trying to approximate a public school curriculum and pace of learning only did a couple hours per day of work on conventional school subjects.
About incentives (bribes). I bribe our dogs so they learn what I want. I use to bribe my kids to go on walks down to the creek. Both with food. I wish someone had given me the idea of bribing my kids to read. I did semi-bribe them by saying I'd buy any book they wanted to read/ look at. "Diary of a Wimpy Kid" got one of my kids hooked on reading that way. I love this review, and fully endorse the idea of doing a non-book review again. Every other year maybe. (I love books too.)
About growing up in an atmosphere that fosters competition.Please do not try to tell me this school does not foster that kind of attitude or that your kids do not partake of it.
<Afternoon programming focused on excelling in “academic competitions” like chess, go, debate, public speaking, robotics, programming and Quiz Bowl.
<This past year four kids placed in the top-8 in a global debate with more than 1000 entries, and two kids are competing at national championships in chess and an academic bee respectively, but not national champions yet.
Come on! Even if your kids don’t get involved in the afternoon “academic competitions” (why put them in quotes? I’d say they’re not “academic competitions” but academic competitions), they are swimming in water with one fat fucking packet of Competitiveness dissolved in it. They’re sucking that solution in through their gills and sending Competitiveness ions up into their brains to have a word with the neuroplasticity supervisor.
I do not doubt at all that a competitive atmosphere has upsides, but I’m here to tell you it has a very substantial downside, too. I’m a psychotherapist and have somehow ended up as one of the go-to CBT therapists in town for super-smart students and young professionals. One person I see was described by the usually down-to-earth neuropsychologist I use for testing as having intellectual powers that are “stratospheric.” There’s somebody who got a perfect score on the LSAT — that’s 99.9 percentile in a test-taking group that’s quite a bit above average intelligence. There’s somebody whose tested IQ when they were high school age was in the 160’s. Unsurprisingly, quite a few of the people are graduate and professional students at the local Ivy.
Some of the supersmarts aren’t Ivy leaguers, though, and were not raised with the expectation that they would be highly successful. And these people come across as having a pretty different take on themselves and on life. There are, for instance, 2 people each of whom had one professional parent and one without a college education. Both were erratic students until they discovered some science and math subjects they loved, then they excelled in those and went on to get degrees in them at places that are not famous, and now have careers in the field. Their point of view about work and achievement is pretty different from the Ivy people’s. They know they are very good at what they do, but they don’t find ways to *signal* that. On the other hand, if you ask them a question that taps into how good they think they are at work skills, they’re cheerfully honest: “Yeah I’m fine at X, Y and Z, but at Q I’m a *monster*! I’ve been playing around with that stuff since I was 16 and I just don’t get stumped by anything anymore. Everybody brings their Q problems to me. And then there’s W. It’s not really in my field, but I was having trouble working with the W guys who did code for some of my projects, so I ended up just learning W myself and doing my own W works out much better. Actually now I’m interested in W itself and have some ideas about a way to revamp it.”
And they don’t seem very prone to jealousy. One works for a company where a management change has brought in a bunch of people who used to work for a couple of the most prestigious tech companies in the world. And he says, “wow. They’re great. They’re some of the smartest and most talented people I’ve ever worked with.” His enthusiasm sounds quite genuine. I can’t tell whether he thinks they’re cleverer than him, but is admiring and entertained, or whether he recognizes that they’re as speedy and clever as he is and are going to be a lot of fun to work with.
Another difference: both have hobbies they love and throw themselves into.
Another difference: neither has much trouble with procrastination.
In contrast, the Ivy Leaguers come across as much more aware, moment-to-moment, of status, whether it’s professional status in the grad program or their job, or who’s making the best quips at dinner. They are vulnerable to intense jealousy, and equally vulnerable to despising themselves for being jealous. They feel deep despair, and might even think about suicide, when forced to realize that somebody is simply more talented than they are.
They have unreasonable expectations of themselves. One became engaged to a woman whose family is mega-wealthy, and felt as though he could not live with himself unless he himself became as rich as the woman’s parents. (He is in a profession where his earnings will make him wealthy, but it will not be wealth on the order of his inlaws’.).
They are very highly motivated to excel, but have trouble with motivation to do the actual work they can excel at.
They are superb networkers, but feel bad about putting more effort into networking than into making new friends and hanging out with their old buddies.
They tend to feel tired and drained after work, and don’t exactly have hobbies — more settings and activities where they socialize and network.
> On Friday the kids are presented with a bunch of locked boxes. One of the boxes has a prize in it. The kids with keys check the boxes to see if their key opens the box with a prize.
Loot boxes. They added loot boxes to kindergarten.
…Okay, fine, that’s not a fair comparison. It’s not a true gambling system because the kids aren’t choosing to spend a resource they could otherwise save up.
But as symbolism goes, it feels like an unflattering metaphor for the system as a whole.
"Once you have freed up half a day for 6-years and a full day for the other seven, you open up a limitless number of possibilities.
Some kids will rush into college classes. Some will choose to use the time to play sports. Some will use the time to master chess or quiz bowl or programming. Some will take time to travel the world with their families."
Sorry to be all grumpy and critical, but I gotta say, that is a *wildly* optimistic and most likely unrepresentative take.
OK, so these kids are limited to those whose parents are both able and willing to spend $40k/year on Alpha, which means. most likely dual-earner couples where both parents have fairly demanding careers. Their kid graduates from Alpha at age 12 instead of 18. Yay! Now what?
"Some kids will rush into college classes." At age 12? Seriously? Are they going to live on campus (in a dorm with a bunch of 18+ year olds? In a frat house?) or are they going to live in their parents' house and commute to and from campus daily, in which case you'd better hope there's good public transit or have a parent who can take you to and from campus daily, because at age 12 you're four years away from being able to get a drivers' license.
Ok, sure, you can do college 100% remotely, but then you miss out on the formative experience of socializing, making friends in person, maybe finding a girlfriend/boyfriend/future spouse.
Some will choose sports! Chess! Programming! Excitement?
Again, how is this going to work *logistically*? Both parents are likely working long hours, probably outside the house. A 12-year-old can be legally left home alone, but cannot drive. Assuming this family lives in the suburbs (= crappy or nonexistent public transit), the kid is stuck home all day. They're not hanging out with other kids their age, because guess what, those kids are at school (Alpha students being a tiny % of the population).
How likely are they to teach themselves chess/programming/other intellectually rewarding pursuits 9 h/day as opposed to, you know, spending the majority of their time on TikTok/video games/p0rn? We're talking *adolescents*, who are not famously known for stellar impulse control, and "intelligent" != "highly motivated autodidact."
Traveling with their family? Sure, that works for maybe 2 weeks/year, i.e., the amount of vacation time the parents have. If you're super lucky, maybe you can send your kid off to travel with a willing grandparent/aunt/uncle for, say, another two weeks, so that covers 1 month out of 12. Also, this works once all your children have graduated from Alpha. If your youngest is still in Alpha or regular school, you can't just leave them behind while you and your firstborn go gallivanting off on a world tour.
TL;DR: This plan has problems, lots and lots of logistical problems.
I haven't seen any comments from Deiseach yet, I eagerly await her appearance with her customary "No shot, Sherlick, of bloody COURSE when you limit yourself to kids whose parents can afford $40k/kid/year for fancy-schmancy schooling, those kids are going to do well! Show me something that works for Tommy from the projects, whose mum is a drug addict living with her fourth boyfriend in the last two years, who also happens to beat her and Tommy."
(No shade on Deiseach; she's excellent at this and I agree with her!)
I liked a lot of this. But whenever I read a long thing that attempts to dig into education research and then cites Bloom's 2 sigma result without qualification, it immediately changes my confidence in the analysis.
https://www.educationnext.org/two-sigma-tutoring-separating-science-fiction-from-science-fact/
This review of Bloom's two-sigma claims by Paul T. von Hippel would itself probably earn a top spot in the ACX review contest. Worth reading. Thanks for posting.
The claim about two-sigma tutoring effect was based on a 3-week study by Bloom's two PhD students.
"... they weren’t obtained by tutoring alone. Instead, Anania and Burke mixed a potent cocktail of interventions that included tutoring; training and coaching in effective instructional practices; extra time; and frequent testing, feedback, and retesting....Bloom’s two-sigma claim had some basis in fact, but it also contained elements of fiction."
I liked this from the article you posted:
“Tutored students received extra testing and feedback. Burke’s and Anania’s two-sigma intervention did involve tutoring, but it also had other features. Perhaps the most important was that tutored students received extra testing and feedback. At the end of each unit, all students took a quiz, but any tutored student who scored below 80 percent (in Anania’s study) or 90 percent (in Burke’s) received feedback and correction on concepts that they had missed. Then the tutored students took a second quiz with new questions—a quiz that students in the whole-class condition never received. If the tutored students still scored below 80 or 90 percent, they got more feedback and another quiz.”
In Made To Stick the authors drive home one of the most important tools for improved performance is regular testing (ie students with 9 quizzes/year outperform students with a midterm and a final on almost all metrics). It seems like Bloom’s tutoring was combined with more regular testing than the control group. I expect the impact he found was a conflation of the combination of (at least) both methods
I don't see how that really negates the point about 1-on-1 instruction accelerating educational performance, though? The point is that a single teacher focused on a single student can give rapid, tailored feedback and quizes, they can't do that for an entire class nearly as easily.
The article corb linked to did not discredit 1:1 tutoring. It just said that Blooms original research was overstated and that a 2-sigma improvement has not been replicated. The article says that stand alone 1:1 tutoring only gets you about a 0.4 or 0.7 SD improvement (still a lot!)
So it’s still very effective and possibly even more effective when combined with regular testing for comprehension.
Main takeaway stays the same, but just caveats on Blooms original claims
Are you saying that the overall cocktail of interventions that Bloom implemented has not replicated the same benefits, or that tutoring alone doesn't deliver the ~2SD increase? Does the overall cocktail of interventions require 2 PhDs per student working 15 hours a day, or something?
Bloom didn't replicate.
Maybe (?) worse is that there is no evidence that the 2-sigma result would generalize or maintain over years.
Doing GREAT for one or two subjects for a few weeks is good. But if that better performance doesn't maintain then ... does it matter?
For what it is worth, I think Bloom's approach was correct. Just that his results are quite narrow.
I recommend reading the book on expertise that this review linked. Having rapid feedback on performance is an absolutely essential part of the "deliberate practice" mentioned here. It's not at all surprising that more frequent feedback works better (although I would still think of something that happens 9 times a year, which is around once per month in a standard school year, as a "unit test" rather than a "quiz").
I don’t love how qualifying Bloom’s 2 sigma has become a shibboleth for being able to talk about education research. But I agree the study has issues, has regressed, etc.
I would say that we hold it as an engineering goal, not a research goal. We believe we can get effects that large and are trying to do so. (I work for this org.)
Most of that criticism, as I read it, isn't relevant to the way the 2-sigma problem is being used here. If you are citing the 2-sigma study to promote after school tutoring, or an AI tutor, or some other tutor-specific intervention, then the details in there matter a lot.
But if you are citing it as a way of saying "we know ways of dramatically increasing performance, but they are too difficult to scale" (as it is being used here), then it mostly doesn't actually matter that they did more than just tutoring, or that it was a special kind of tutoring, etc. If anything that strengthens the point, that those results are _even_ harder and more expensive to scale.
The fact that it was on a narrow set of tests is somewhat more convincing, but, at least for me, doesn't really do much to change the larger point. I think adding those caveats, in this context specifically, would have just been extra cruft (on an already long essay) that wouldn't have changed the message at all.
I would love to hear a perspective from another parent from the acx community to see if their parenting ties out. It strikes me as almost too good to be true.
FWIW, some relevant Reddit threads with negative reviews are
https://www.reddit.com/r/Austin/comments/17feoeu/a_school_with_no_teachers_alpha_private_school_in/ (mentioned in the OP) and https://www.reddit.com/r/Austin/comments/1hbahyt/private_school_comparison_alpha_versus_magellan/
Yeah… I’ve seen this. The commenters are all Reddit-flavored criticism from people who have no direct experience except for one (“dipnoob”), who said his daughter didn’t like it after going for a year. In his post history he also says he has an 18 year old trans daughter, and given the schools association with Elon musk it seems like that could easily be an area of conflict that isn’t necessarily related to education quality for the average student.
There are also two comments from actual students at the first link (ProgressPlane4465 and Relevant_Habit6774), both have no other reddit history which may or may not be suspicious (I guess it's easy to sign up just to comment on a single post and then it gives you this sort of random username? But also someone who wanted to astroturf would just do the same)
"I guess it's easy to sign up just to comment on a single post and then it gives you this sort of random username"
Yep, reddit selects a username for you
One of the comments on the second posts says
> I know a family who goes to Alpha. They don’t learn history, and the other things I’ve gathered from the family about the school make me certain I would never send my kid there.
Its from 7 months ago so I wonder if they have talked to this family since 2022.
Links to two comments from kids who go to the schools
https://www.reddit.com/r/Austin/comments/17feoeu/comment/mr5i99q/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button
https://www.reddit.com/r/Austin/comments/17feoeu/comment/lm1ezva/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button
Hi Nicholas! I have had 2 students at Alpha's flagship location here in south Austin these past 3 years and have been blown away by their experience. Not usually active in ACX comments but have been reading for years and my husband is a prominent writer in this social neighborhood broadly (Byrne Hobart of The Diff, https://x.com/ByrneHobart/ )
My 9yo daughter is quite high ability and I very much doubt that anywhere else would have let her move quickly. I never had to make a single request for accelerated material or schedule a single conference to complain that school was too easy or boring for her, I have basically no friends in Austin who have not faced these issues in various other charter and private schools for their children. She has maintained the 99th ability not just for NWEA MAP score achievement but also the top few percentiles for growth throughout these 3 years.
My 7yo daughter is also bright but a tough cookie, the label is "2e/twice exceptional" i.e. on the spectrum but also gifted. She spends very little time on the apps because the school environment is challenging for her. Her growth has also nonetheless been incredible - I am extremely confident that the academic environment Alpha provides via its 2 Hour Learning platform is the most potent available today. https://x.com/gtmom/status/1931058079745056974
I believe in the model so much I've gone to work for Alpha/GT school and am moving my daughters to the GT school this fall.
Wow, thanks for this. Another data point that makes it sound amazing.
I am a parent at GT School, the gifted school of Alpha. My son, Nate is reading at a 7th grade level, is in Pre-Algebra, and similarly ahead in science and language. He is 10 years old, and a young 10 with his birthday mid-April. After being bullied in two different (well-renowned) public ISDs we chose to switch to private. His first week of school he was beaming and exclaiming that kids “got him” at this school. Alpha truly does foster a love of learning and does accelerate learning. Though I agree it may not be for everyone, my son loves it and is thriving.
Also he is testing in the high 90th percentiles 96-99 across the board through MAP
I'm one of the parents at the school and posted my perspective in the comments.
The part about kids hounding you about wasting water was familiar. My 4 year old daughter is always like, dada turn off the water you don't need it :)
https://www.reddit.com/r/homeschool/comments/x86xfq/how_long_do_you_homeschool_per_day/
Seems like most homeschoolers are already spending way less time than they would in a traditional public school, so this piece overstates the impressiveness of Alpha's homeschool pilot results.
Second this. It’s nice that it’s preassembled, but my kids are homeschooled and advancing at 1x speed on less than 2 hours a day (for a lot less than $10k per year).
I always point out that Finland and South Korea get near-identical PISA scores despite vast differences in hours worked, and Scott already did a piece arguing that homeschooled kids who do little or no work are only about a year behind their public-school peers.
https://astralcodexten.substack.com/p/kids-can-recover-from-missing-even
That said, learning 2-3x faster with only 2 hours a day is pretty impressive, for the full Alpha program. It's a shame the entire progressive establishment are impervious to evidence.
"That said, learning 2-3x faster"
I haven't read the entire review (yet) but 2-3x sustained would have the kids performing at college freshman level by about grade 5. Do they? Or does the 2-3x mean something else?
I think the article mentions top-end students performing at the level of high-school graduates by grade 3, so... conceivably, yes? Maybe you hit a point of diminishing returns at some stage once you get into adolescence/adulthood- there are a lot of interventions that give impressive results in kids that don't persist into adulthood, although some socialising interventions do seem to be longer-lasting (I think Charles Murray wrote about this at some point.)
But even if the only permanent benefit is *not* squandering vast amounts of childrens' time that's well worth pursuing. Bring back child labour, unironically.
"I think the article mentions top-end students performing at the level of high-school graduates by grade 3, so... conceivably, yes?"
So I'd then want to see how they define "level of high-school graduates".
Do these 3rd graders understand Newtonian mechanics? Can they read, day, the Fagles translation of the Iliad and understand it?
Or does this mean that they can read a bit beyond Harry Potter and reliably work with fractions?
Do you know where to find the definition?
I'd... recommend reading the whole article and maybe contacting the Alpha admissions department, then?
So this is mainly measured by MAP test scores, which heavily correlate to SAT/ACT scores. This is mainly going to measure reading and math proficiency. So my guess would be, yea, a fifth grader in this system would probably be able to read and understand Fagles’ Iliad (though the depth of understanding is going to be limited by emotional development). Having the reading ability would enable them to have at least a conceptual understanding of Newtonian physics if they’ve encountered it, and I would think that if the curriculum contains instruction on application of Newtonian physics there’s no reason they can’t understand that as well.
It's in the review. It's entirely MAP score based. So top level 3rd graders get the same MAP score (multiple choice testing) as median high schoolers. So they probably don't understand Newtonian physics or read the Illiad....because median high schoolers can't do those things either.
(I work for Alpha.) We don’t think our homeschooling platform is impressive (yet). And yeah, “you can use time more efficiently than schools for learning school things” is not magic or unique to us.
It’s still true though, and I’m happy to be working on a project that elevates it to a constraint in the school system.
This right here. Homeschoolers often spend way less time per day and get good results, with a wide spectrum of curriculum and approaches. That these Alpha folks did something similar and got results like that should not be news.
Loved this.
This is really interesting and will directly change my parenting from this day forward: my husband and I spend 30 minutes every morning 1-1 tutoring our kids, and I will start adding financial incentives from next week. Our 3yo can read and write (not very well! But better than most 3yo!), and the 6 yo is very good at reading but not quite at adult level fluency yet. I'll be paying him to read the Roald Dahl corpus and for completing handwriting worksheets. The 3 yo can get money for finishing each phonics book. Thank you for an informative review!
Any tips on introducing early reading and writing? I have a 2.5 year old and I’m very encouraged by these non traditional models. Big fan of 1:1 tutoring. Thanks!
There are two approaches I've used. The first is Siegfried Englemann's famous "Teach your child to read in 100 easy lessons". The lessons are not easy and it took us 400 days to complete the whole book with my first son. However, it was successful and my son did learn to read. It was a painful process and the Distar orthography was somewhat frustrating and at odds with the Phonics approach taught in his reception class. Fortunately we live in England which has fully adopted evidence-based reading education. I didn't do the writing element which I regretted - my son's handwriting and writing ability was far behind his reading level which held him back somewhat. I will be teaching him "joined up" handwriting this summer before Year 2.
With my younger son we decided to use a different approach. We bought Read Write Inc flashcards (the phonics scheme his eventual school uses) and taught him the Set 1 sounds. We started "blending" very early, but this is the part that took a huge amount of time and perseverance for both my sons to be good at. This is going from c-a-t to "cat". I.e. blending "c" and "a" together to get "ca". We worked for 3 months on this before he got it - we basically did direct instruction (e.g. going "c-a, ca" now you try - the copying works!).
We then bought a different phonics scheme books, so that when he starts Reception (this September - he'll be a summer born kid - we're not holding him back - I still question our decision) he won't be bored/remembering the books so they can properly assess his level. He started reading these - exceptionally painfully - but it worked. I then introduced writing, again using RWI - first using wipe away sheets, then buying the RWI workbooks. I also make him write his name every day - he's a pro at that now. He writes something every day (i.e. The cat is is on the mat).
We've now moved onto Set 2/Set 3 sounds. And we've memorised high frequency non-decodable words, as well as working on sounding things out in your head.
It's pretty painful at times but very rewarding to see his progress.
How do you prevent them from just memorizing the phonics books, and instead actually learn to sound them out? I feel like working through sounding out with my 4 year old, she'll struggle through it to get a word, but then she just kind of knows it, and doesn't have to sound it out so much the second time and I feel like she's not actually getting good practice with reading but just partial memorization of what the word looks like.
You do both. We have "speedy green words" which we get him to sound out in his head and just "say" and harder words he has to sound out. You sound like your kid is already getting a great level of semantic knowledge where they don't need to sound i tout - lucky you - mine laboriously sounds out every word (the younger one) apart from the non decodable memorised words.
Yeah I know it's sort of a good thing, but we can't use the same texts too many times to practice sounding out so she's acquiring words but not getting that much better at sounding things out. I guess it will come, and we just have to get new books a lot.
We use little wandle letters and sounds, they have 9+ sets of books with 6-12 books per pack. not in danger of running out any time soon!
OK have reread this. We don't have this issue as the phonics scheme we are using has a huge number of books. We reread each book twice and then move on. There are over 100 decodable books in the scheme.
When my wife and I were teaching our child to read we had "hold out" books (maybe a Clifford book or a Little Critter book) that we knew the child had never read before. When the child could read one of these, we'd score his as reading.
This did NOT prevent him from simply memorizing all the words that would appear in the book from exposure in other books, but that is what we did. The "if he can read a brand new book" metric seemed to track what we meant by reading.
IME it's mostly genetic. My first learned all his letters by 18 months. My second still didn't know them by 5y. I didn't do anything differently; in fact I did so much more with the second because I was concerned. Turns out the first one was autistic and hyperlexia is a symptom. More of a curse than a blessing, in the end. Meanwhile the second did eventually catch up.
I was convinced by this review's defense of incentives, but I *don't* think paying a kid to read Roald Dahl books in particular is a good idea. Reading fiction is something you want kids to enjoy for its own sake, and I think there *is* convincing research supporting the view that rewards undermine that sort of intrinsic motivation. (See https://www.alfiekohn.org/article/newt-gingrichs-reading-plan/ for a brief overview.)
I think rewarding kids for completing educational modules is different, because we don't really care whether kids enjoy those things for their own sake--the educational modules are obviously just a means to the end of acquiring the necessary skills and knowledge. But if you want the kid to WANT to enjoy reading for its own sake as an adult, paying him to do it now is counterproductive.
Isn't the whole point that you give them £1 to read Roald Dahl books and it doesn't impact on their enjoyment at all - instead it may actually improve their motivation to read even without monetary incentive!
I can't see how paying my son to read a great book would make him less likely to enjoy it.
I also don't particularly care about intrinsic motivation. I don't see why anyone should - I'm familiar with Alfie Kohn and think his influence on education is malign.
I think in a perfect world, kids would read a few great children’s books and spend their free time doing other things than reading.
But my experience of kids myself included and, coincidentally, my genetically related kid - who go on to become constant readers as adolescents and adults, is the personal search for books, almost furtive and almost entirely undisturbed; a lot of which is plowing through a fair amount of garbage, or short of that at the very least, deeply unedifying books.
Parents should probably not try to make readers.
I spent my whole childhood reading - it was an immense source of pleasure for me. I hope it will be for my kids too. But why on earth shouldn't we try to make readers, given the links between volume of reading and educational outcomes?
I guess we need Scott A. to craft a survey question to get at: if you are a reader - someone who, at the thought of a threat to vision, as is beginning to happen at my age, of course I’m nearsighted and astigmatic but I don’t mean that - your second thought is not: to lose the sight of beautiful waterfalls, or relatives - but rather: the fear of not being able to read text many hours a day!! - if you are that …
Were you made by your parents? Would you not be so without parental involvement? Did your parents have anything to do with this at all, or was it more like a pulling away from them into a private world?
I don’t know many readers (any beyond the nuclear family?) so they may have been generated by those contests in school or parents saying, now is your 30
minutes you must spend reading.
Some of us did indeed get in trouble for reading too late; and I myself once took my son’s copies of LoTR and hid them away.
I developed no useful skills as a kid but it may not be fair to blame reading. I watched a good deal of TV as well - enough to be full up for my whole life as it turned out.
This is a very interesting take I've never seen before - what's the negative of reading? That many books are bad? What would you want kids to do instead?
I am a lifelong voracious reader. It is a lonesome activity and my social skills are certainly worse than if I read 20 per cent less.
I can see why common-sensically it might sound silly that paying a kid to read would make them less likely to enjoy it, but there does seem to be a fair amount of evidence that it can genuinely have that effect.
For example, there are experiments that show that if you reward kids for doing a particular activity on one day, they're less likely to freely choose to do that activity if given the option on the next day. Now, admittedly, we have to take basically all psychology studies with a grain of salt, but in this case, I think this really does line up with my own experience of my own motivation. If I got paid to do something before, but then later I'm merely offered the opportunity to do it without pay, I have some tendency to now think, "Why should I do it if I'm not getting paid?" Doing it "for free" now strikes you as a waste of time. And this does seem to be born out by research.
Sending the message that reading Roald Dahl is something you need to be bribed to do, rather than a delightful thing you *get* to do, seems like a bad idea to me.
But if you don't particularly care about intrinsic motivation, I guess we might stand too far apart to have common ground for conversation. I can't imagine not caring about intrinsic motivation. To me, it seems that happiness in life absolutely depends on being able to enjoy for their own sake as many of the things you *have* to do as possible.
I think all the evidence we have shows that intrinsic motivation is one of the most heritable things there is! So to the extent my children do or don't have it, that's not something I can influence.
The research quoted in the article seems at odds with what you're saying - that actually rewards don't make kids less likely to pick that activity - they make them more likely to choose it!
There’s a big difference between an incentive to get over a hump and kickstart a process, and an incentive as a crutch to solve for an exercise that a child really and enduringly does not want to do.
If my picky toddler won’t try a donut (a real example), I can bribe them to take one bite, and then the issue is obviated. This kind of thing comes up all the time. The point is not to be an imperialist or a reductionist about incentives, it’s to have them in the toolkit.
My husband gives our kids money for household chores (which I disagreed with - I felt it should be an expectation not something to he rewarded). He's read my comments, disagreed with bribing to read books and has countered with incentives to use family Duolingo which I have agreed with.
Not sure why that should sound silly. After all it’s a pretty common observation among adults who make a beloved hobby their career is that they lose a beloved hobby.
Of course mileage can and does vary there – might also be an interesting thing to study: What about personality and activity structure makes adding material incentives to passion work or fail?
I say, don't micromanage. There aren't that many Dahl books. Let them read what they want and pay up. As a kid I hated when something that sounded fun turned out to have all these terms and conditions. "Good practice for being an adult", yes. "Incentive", no.
That's a good idea actually. Read any book of x pages or more and get a quid. I like it!
The point of the argument here is that external incentives is the first step on the ladder to internal motivation.
The hypothetical for reading fiction would be:
Child doesn't like (or maybe just doesn't care about) reading -> incentivized to read -> finds out that some books were actually enjoyable -> decides to keep reading on their own
That, to me, is at least plausible, as a way of encouraging a child to find out whether or not they might enjoy reading. For it to work, you probably want to be incentivizing a wide range of different books until they find out what they like.
There's also the skill hump to get over. When you're not yet that good at reading, reading a whole short book is a lot of effort, and the story isn't enjoyable because the struggle of actually reading it is so hard. After a lot of practice, you get better at reading and it becomes easy, and the story is its own reward.
I'm still undecided on the idea of monetary rewards for basic tasks, though. I know my kids will do just about anything for a dollar, but I also know that if I overuse this then soon they'll be refusing to do anything without getting paid... and then they'll start negotiating for more money.
Curious: do you think *monetary* rewards are the best way to go with a 3yo? Back when I was bribing a 3yo to read, it earned him ice cream; I'm curious about the relative merits of ice cream vs cash. (That former 3yo is now 6, and I'm planning to bribe more reading with movie nights -- TBD how well that works.)
I think there is some evidence bribing with food might lead to a bad association with food = rewards = obesity. I don't know how right that is, but in any case, cash rules all and you can use cash to buy what you want, including an ice cream. I think he's probably too small to get it just yet, but we'll see.
My approach was "the 3yo should get an immediate reward; an extra dessert seems like a sustainable level of immediate rewarding". (This is presumably the mechanism behind stickers, I just never quite figured out what kids are supposed to *do* with the stickers they get, and consequently my kids don't quite seem to know either.) If cash works as an immediate reward for a 3yo (and if you actually keep cash in small denominations on hand! -- part of my problem was that I don't), then I don't disagree that it's equally good.
I will report back!
This is really interesting, thank you ! As a parent of similarly aged kids, do you have any advice on how to make those 30 mn a day a pleasant time ? I feel that every time I try to teach my son, I have a hard time staying calm and he is not very interested either.
Stay calm, over praise your child for tiny successes ("oh wow!!!! you're soooo clever - that's right - that sound is "oy, a toy to enjoy"!). Be very firm; the exercise/reading does not end until you say it ends. Use rewards/incentives - we already say "you can watch TV, but first, finish your reading". Be very clear at the beginning what the expectation is - i.e. "first we're going to practice your writing with the yellow sheet, then we'll do your red book, then you'll write your name, then we'll practice some sounds, then we'll do red words, then we'll read two pages of your book".
As someone who doesn't mind breaking conformity but DOES mind 40k a year, is there any other way to get something similar to their software?
I think the home school product has become available. But my googling can’t find it. It’s been quoted at $10,000/year though and doesn’t include afternoon programming, “babysitting” or incentives
You can get on the list to be notified about the homeschooling version of 2HL here https://go.alpha.school/home-school
In Alpha's pilot of this program, however, it has been much harder to get the 2x+ learning results in the home environment without the wonderful guides and various tailored incentives.
Thanks! Surely you could do your own incentives program if you can see the testing scores?
Oh certainly, it's just that homeschool users (even early adopters) may not understand how critical the motivational model/incentives really are. We can probably do more in that regard to educate/support as we continue to experiment and roll it out, but ultimately Alpha cannot control whatever is happening not within its walls.
I homeschooled another child of mine for a while and it did not go well for many reasons but I did find that incentives at home felt much more manipulative to him - whereas learning alongside a bunch of other kids who are also racking up their app time and cheering each other on and sometimes going on outings together as a reward feels way more natural. At home with just me and him it felt admittedly lame : "do your work!! don't you want these stickers?? chop chop " 🥴
Love that point at the end! I think eventually some kind of cohort-driven model with the homeschool program so kids have that would work well.
I wonder why their marketing leans so heavily on AI instruction when that angle is very unpopular and seems not even really true?
Is it unpopular among their current target audience?
(Alpha parent who now works for them) There is AI behind the scenes refining our apps platform, including eye tracking and lots of interpretation of student learning behavior to determine when the material needs revision vs. when the students are choosing to waste time, etc.
But you're correct that the "AI tutor" is not like talking to an LLM/chatbot or anything of the sort, and that many people who see various media coverage are misinterpreting it as such.
I wasn't part of that strategic decision, to lean into "AI" branding but, as a parent who was considering Alpha a few years ago, the "AI" component didn't loom large in my own decision making. It has probably changed by now as AI enters the popular discourse more and more.
I feel like "data driven" would communicate the same thing in a snappy way without sounding like "we put your kids in front of a chatbot."
Given the section about their current consumer base, it might be that it's really unpopular among the general populace (although even that isn't actually obvious to me) and much more popular/less unpopular among the group of people who are considering these kinds of schools.
If that's the case, it might make sense now, but it could wind up being a poor choice down the road depending on how large their scaling ambitions are.
I believe it's for PR & virality purposes. Everyone's wondering how AI will "transform education" and journalists are hungry to write about it. Alpha Schools has existed for a while, has always used personalized apps, and only got this level of media traction when they started saying it was 'replacing teachers with AI.'
If that makes some parents and educators angry, all the better for provoking discussion.
I mean, there’s also a sense in which it’s true. The in person teachers don’t give academic instruction or do lesson planning. AI is integrated into evaluations of mastery and decisions about next steps.
But yeah, I do think that people are (naturally and understandably) misreading it as “replacing teachers with chatbots”.
Hi Matt! Definitely agree there's a sense in which it's true - though the guides give some form of instruction during afternoon workshops, yes? And their presence (paired with incentives) seems to meaningfully drive completion of the 2 hour learning, given that homeschool students achieve less impressive results. And the virtual coaches available for extra support when kids get stuck could also somewhat qualify as teachers, right?
And yes, although I'm pretty deep in this stuff (spoke to people at Alpha back in 2020 before Gen AI boom) I was surprised to read that chatbots aren't a bigger part of the 2-hour-learning model now.
I think when people hear about AI replacing teachers, they expect this to mean that education is being done cheaper and more scalably - whereas Alpha's pitch is AI = education done faster + more effectively. It's more like "AI replacing some of the typical functions of a teacher's role, while actually having a student:teacher ratio lower than typical schools and paying teachers more, leading to a counterintuitively higher tuition price." Of course this is addressed really well in the piece, but I agree there are major scaling concerns. Nonetheless I do think there are ways to make the model work to some degree at home + virtually.
Gonna try to find time to write a higher-level comment on the pedagogy since this review is so rich and fascinating. For full disclosure in this thread, I've spent the past 5 years building Prisma, a virtual school in the same innovative education space.
Yes to all of this.
An enormous portion of our costs right now are paying for AI models, which will come down.
Teacher salary is actually not the bottleneck, which I think should remain high. Nor is teacher:student ratio, which will be and in some cases is higher in other models that use the platform successfully.
The system does not function without teachers and the purpose of the system is not to replace teachers. The purpose of AI is specifically perfect and automate academics, which is a process that is not complete for some of the reasons you note.
Edit: I think the reality is that we are just dipping our toes into scale. There’s a lot to figure out. That’s the attitude from this side of the project.
Love all that! And I think that messaging will resonate and not offend - most parents know and feel that the best teachers are the best at driving relationships that enhance learning, and academics can be optimized other ways.
Hi Emily! I am Alpha parent who now works with them/Matt as well. You're right that there are several sliders here, not far from the usual "good, fast, or cheap" pick two situation. Unfortunately conventional education actually picks 0.
The counterintuitively high tuition price has several causes: huge expense of the behind-the-scenes tech stuff (building and improving the apps, this is AI powered with eye tracking data etc), guides who earn 2-3x public school teachers in the same areas with dramatically higher teacher:student ratio, and some of the workshops are quite expensive (indoor climbing 3 days/week, swim lessons, field trips to a ranch...). These could be adjusted in various ways in the future , like well paid guides in different ratio, cost effective workshops, cheaper buildings in lower cost zip codes.
The guides do create and implement the afternoon workshops, and some of those have used LLM/chatbot type AI, as a kind of educational dessert and not the main course. Here is a video about the recent "AI Imagineers" workshop, my daughter is the one making animal trading cards. https://www.youtube.com/watch?v=foGBJHdjhZg
Will look forward to your additional comments!
Very cool, love the video. Can easily see the way 2-hour learning could become the seed/base and some models build on top of it in a much cheaper way.
This sounds very cool, though I neither live in Austin nor have a spouse who is open to homeschooling, so I don't think I'll ever be able to use it. Alas.
The one thing I *wouldn't* be happy with is the lack of a few things that I considered really essential to my own education: (a) significant periods of sustained silent reading for pleasure in elementary school, (b) in middle/high school, classes where everyone reads a classic work of fiction over a few weeks, discussing it as a group as they go, and (c) lots of practice writing, including creative writing, where it gets read and critiqued by a human--at least by a human teacher, but sometimes also by fellow students.
These are things that you can't (or certainly shouldn't) get AI to grade, where performance isn't going to be clearly quantifiable, but they should be included.
I had similar thoughts! I’d heard that kids in general aren’t reading full books in school these days, so if there were a way to combine that with the accelerated study in other areas, that would be ideal…
I went to a Montessori school from preschool through 6th grade, so I have a lot of appreciation for parts of this approach (I got to have a much more self-paced experience than most kids do, though obviously not to this extent).
The things I would probably not have liked:
-So, if you’re in the GT program, you have to do chess *and* quizbowl *and* public speaking? Sounds great if you’re into all of those things; not if you’re interested in some but not others. Or is the idea “some schools choose to include Latin or programming in their Official Curriculum; we’re choosing to include public speaking in ours, so everyone has to do that”?
-Similarly, I have this vision of my 7-year-old self cruising through 4th- or 5th-grade math and English while still officially not having “leveled up” from first grade due to not doing, say, the paper airplane thing (maybe due to some difficulty at first, subsequently out of something like spite).
-Not a fan of the idea of everybody learning just enough piano to play a grade 1 solo, and then not playing piano anymore. Maybe this is just my personal bias (for me, chess is the thing I would have chosen to pick up for a few weeks and then put down, but for them, it’s music).
Hi there, I am Alpha parent who's gone to work for them, my 2 daughters are moving to the GT version this fall.
Since core academic learning is so efficient when done on the Alpha apps platform, there is a ton of time for workshops, even with a relatively short school day that starts late (~8:45a-3:30pm). You're right that some kids will like some of those activities more than others, but they've got to be doing something constructive with the time, and the workshops rotate each short session (~6 weeks) so it's not like being stuck coming into school at 6am every day for quiz bowl that you hate.
I have exactly that kind of salty first grader lol (cf https://x.com/gtmom/status/1931058079745056974 ) . In practice the wonderful guides eventually get them through each of the check chart activities required to advance levels, though students do genuinely do this on different schedules and may indeed be delayed for failure/noncooperation. Students can move up at each session throughout the year so this is not disastrous like being held back an entire year.
Since GT school is about to go into only its second year, we are still experimenting with the schedule and may be moving some of those activities into an optional after school hour for more depth and continuity.
Personally I hope they/we also find a way to do piano (or music broadly) more continuously, my daughters missed that one! stay tuned!
I’m planning on adding literature (and history and some other things) along these lines to the Alpha system. There’s no resistance to it, and there are patchwork solutions in place in the interim. It’s just a young project.
>in practice they use 20:1 guide:student ratios (vs the 5:1 ratio at the Alpha private schools)
These ratios are reversed, right? Or do they really have more teachers than students?
There are some ratios later in the piece with the same problem, and since they refer to normal public schools it looks like they must be all reversed. Is it in the spirit of the contest to let the author fix errors like that? It's pretty distracting.
Yes those are reversed, fewer teachers than students (I am Alpha parent)
They're reversed, as 4gravitons explains. There's only one place where it's stated correctly:
"At the five‑to‑one student‑to‑teacher ratio Alpha runs[...]"
The review also suffers from bizarre use of hyphenation, misuse of homophones and near-homophones ("skimming" vs. "skimping", "shoot" vs. "chute", etc.), and other MUGS problems.
The review is very informative and extremely thought-provoking, but I expected that a finalist of an essay contest would have better writing fundamentals.
I “disaglee”
> "When the data and anecdotes disagree, the anecdotes are usually right. It’s usually not that the data is being miscollected. It’s usually that you’re not measuring the right thing."
I'd like to offer an anecdote of a recent Alpha School graduate I know that supports this review (which was great btw).
I know a guy who graduated from Alpha School in ~2023. The class size was only about 10 people, so he was either a late entry, or one of the early beta testers from back when Alpha school was a new concept (thus likely not getting the full benefits of K-12 Schooling). This guy is one of the most impressive people I've met.
In his senior year, he raised ~$3 Million dollars to build a dirt bike park outside of Austin, apparently with some competitive advantage of shuttling bikers to the top of the hill and being larger than the other parks. He successfully built the park, operates it at a profit, and is now building a national mountain bike apparel brand that seems to be doing quite well. He's 19 by the way.
He told me about Alpha school shortly after graduating when he was looking for his next cofounder for his new company, and I remember I was thoroughly impressed at the time. We were both looking for cofounders at the time, so we had some in-depth discussion on past achievements, business philosophy and whatnot, meaning I got a good look into how he had become so successful literally in high school. It didn't end up working out, since we were both guys in the "driving seat" so to speak, looking for people to join our existing venture and business idea rather than looking to join someone else's.
He obviously isn't going to college, but he doesn't need to since he owns the majority of a company that returns more money than the average American makes in a year, while also owning the majority of a $2M+ piece of land. If he wanted, he could probably not work for the rest of his life, and he's 19 mind you.
Now you might be thinking; "Ok. This guy is a rich kid. His dad or family friend probably gave him millions of dollars and he built a return-producing asset with it, big whoop. It's no big deal to build a known business model with daddy's credit card." and that's the same thing I was asking him. It's impressive to build a company, but a lot less impressive to replicate a known business model when given a large amount of money to throw at the problem. Bike parks are certainly common enough, and it's not like he was building the new "Uber for dogs" so with a few million dollars it seems like no surprise he succeeded.
How he raised the money is the interesting part. He cold-called (actually called, something people don't do so often anymore) every multi-millionaire in Austin he could identify until he found one that would listen to him. He pitched the idea; secure $1M in investment, purchase a piece of land outside of Austin that's a good fit for a bike park under a mortgage, operate the bike park (and apparently there was a lack of bike parks around Austin) according to known business principles, and return a profit. He was a competitive mountain biker, and had worked at a bike park previously (I assume one of the afternoon activities that Alpha School encourages), and that was a very attractive part of the pitch to an investors.
Anyway, to sum it up he raised money, bought a property, built his bike park, and is now building his next thing. I've met a LOT of people in the startup space, like literally 500+ founders, and his story sticks out to me as one of the most practicable (he didn't try to reinvent the internet in his first company) and successful (he actually built a profitable company on his first attempt). Now you could say, as a lot of people like to do about successful startup founders, that his success was due to his unique position. That it wasn't his individual magic that led to success, just his circumstances (Bezos raised money from his parents, Elon's Dad's Emerald Mines, Bill Gates being the 0.01% of children with regular access to a computer when that was uncommon, etc.) but... isn't that kind of the point here? If that's the critique we levy against him, then that's an accolade for Alpha School! Since they're the ones who created the circumstances for him to succeed in a major way.
I don't have kids, and probably won't for the next few years at minimum, but when I do, Alpha School has been at the top of my places to send them to school for the past couple of years. My previous plan was to do elite Aristotelean-like home schooling/tutoring, possibly in conjunction with other like-minded parents, but if there's a functional model already out there, why reinvent the wheel? This review simply increased my confidence in Alpha School.
My prediction is that this review will win. SSC readers already are really interested in alternative forms of schooling, it's very well written, extremely detailed, and probably the most "review with potential real world impact" on the list of finalists.
The only bad thing I have to say about Alpha School is the branding. A-priori, the name sounds like something Andrew Tate would come up with, and its relative unspecific "We use AI to teach kids to learn!" copy on their website makes them look like the very many AI vaporware startups and/or Web3/Crypto/NFT bs. This is completely a personal issue with their branding, not substance though, and is just my personal impression, and I don't think they should change their vibe because of the association with less savory lines of business.
Edit: Apparently he is going to college. When I last had a long conversation with him it was summer after his graduation (or maybe he was intending to take a gap year? I don't remember), but apparently that changed. It still stands that he, more than most people didn't HAVE to go to college though.
Great comment.
I think the kid was Rhett Jones. He didn’t want to go to university but his parents insisted he go. I think he ended up choosing University of Austin (not UT Austin) as they agreed to give him some credit for running the bike park.
(I would love to read a review on the University of Austin from a ACX reviewer…)
Also worth noting that Rhett is definitely to most successful to come out of the Alpha High program. He was in the first class and he is the one the school brags about the most (they even filmed a documentary - I think this might be it — https://youtu.be/IsUwcO61irw?si=TkObs_CY34iwD2Pj )
Ahahah I wasn't aware he was their biggest success, and that definitely qualifies my anecdote. My assumption was that while he was better than average, doing great interesting things was pretty typical in the senior year at Alpha School, although I now realize that might not be the case.
I wasn't aware he ended up going to UATX, but it makes sense. It's probably unnecessary in terms of educational outcomes, but university provides structure at a time in life when structure is quite important. Wunderkind with money, prestige and free-reign do whatever he wants is a very common backstory behind washed-up adults. Having the skills and mindset to build a business is rare, and incredibly valuable, but that isn't necessarily the same skillset for a life of flourishing. Especially if you lack some other important things like humility, genuine friendships, discipline, delayed gratification, etc, which quick success early in life might make harder to develop or retain.
From my interactions he seems like a really great person though, although I don't know him personally well enough to make any judgements. Life is long, and 3 years at a unique university won't set you back if you're already well ahead of the pack.
Thanks for sharing your experience. (I'm not a paid subscriber, so I can't upvote.)
Likes are turned off for everyone including paid subscribers (This is unique to Scott on substack). If you read on mobile you can still like comments, but that's not intentional.
Interesting. A week or so ago, I was notified that someone upvoted one of my comments on ACX. I assumed the distinction was between paid and unpaid subscribers; it's funny that it's actually because of a bug in the mobile version.
>Caplan showed that, within the normal range, nothing you do in education or parenting matters.
If I am not mistaken, didn't Caplan argue only that nothing we do in HIGHER education matters? Has he made that same claim re K-12? The decline in test scores associated with COVID remote learning (and decline in attendance) certainly implies that K-12 is more than just signaling,
https://www.nationsreportcard.gov/reports/mathematics/2024/g4_8/national-trends/?grade=8
https://www.brookings.edu/articles/student-level-attendance-patterns-show-depth-breadth-and-persistence-of-post-pandemic-absenteeism/
Bryan addresses that directly back in 2015:
That is remarkably unconvincing, since the only "data" he cites is his n=1 personal experience. And is completely inane:
> My K-12 memories include thousands of hours studying material I knew I’d never use after the final exam. The three years in Spanish were especially traumatic, but they’re only the tip of the signaling iceberg. How many years did you study poetry, art, music, history, and civics during elementary, middle, and high school? Even math often smells of signaling; honors math students usually have jobs that don’t use advanced math.
The test of the effectiveness of three years of K-12 Spanish is not whether the student ends up happening to use it; it is how proficient the student is in Spanish.
And the trope of "I will never use it, there it is valueless" is completely inane. First, although I have not used 90 percent of what I learned in school and you have not used 90 percent, IT IS NOT NECESSARILY THE SAME 90 PERCENT. A school cannot know what each student will use in the future. Second, the idea, because I do not use historical knowledge at work, the history I have learned is useless is obvious nonsense. Eg: How do know that "the Democrats are the real racists because slavery" is a dishonest claim? Because I know who the Dixiecrats were, among other things. How did I know that the claim (which I heard on NPR at the time), "the Obama Iran nuclear deal is really about oil, because the Secretary of Energy is involved in the negotiations" is stupid? Because I know enough civics to know who the Secy of Energy was. How do I know that, no, the chance of Biden winning PA and GA was not one in a quadriilliion? Because I know enough statistics to know that what assumptions underlie that claim.
And, I note Caplan, and economist does not bemoan the hours spent teaching Economics, a graduation requirement in California and, I assume, elsewhere. How odd, since almost no high school grads use econ at work. But some economics knowledge sure comes in handy in deciding who to vote for for mayor of New York.
Honestly, Caplan's take here is no obtuse that it is hard to take anything he says without a very, very large grain of salt. He certainly should not be cited in an article re K-12 education.
Bryan addressed the confusion about his beliefs about k12 directly back in 2015:
https://www.econlib.org/archives/2015/02/signaling_in_k-
(He may have changed his beliefs since then)
Your link is broken, this is the correct one https://www.econlib.org/archives/2015/02/signaling_in_k-.html
On Caplan's beliefs, I believe he has said basic reading, writing & arithmetic can be learned early on in school, but after that it shifts away from building human capital.
Quick note on a minor confusing thing: This post at one point says that a "mastery test" is like a final exam, but later implies that it's something that happens every day. I assume this is just a terminological error or something.
HI Taymon, I am Alpha parent who's gone to work for them. While the mastery tests are indeed like final exams, the other mastery implication probably refers to the fact that, day to day, Alpha's apps platform enforces mastery learning on each discrete skill.
Mastery learning is a core Alpha value. Most (possibly all) commercial education apps out of the box do not insist on mastery and instead pass students along with traditional "passing" scores of ~80%, which soon leaves them with big holes in their knowledge.
To be clear, I was not making a comment on Alpha School itself, I was just making an editorial suggestion for how the author of the review could make it slightly clearer.
Understood, just for the drive by other readers out there, cheers!
Our family has found that incentives work, and video games are a particularly effective carrot. Our homeschooled kids have to complete their modest IXL workload and practice piano for 30 minutes to unlock 30 minutes of game time each day. They do it pretty much automatically each day with minimal complaints.
How did you develop your homeschool curriculum? Having a kid soon and would like to homeschool but don’t know where to start
We're pretty relaxed. We figure if we can ensure our kids are literate and numerate, they'll be able to learn whatever else they ever need to. Of course, they do more than reading, writing, and math, but that's our core focus.
We used to use Kumon workbooks, but I was manually grading them so sporadically that they didn't get meaningful feedback. That's where IXL really shines--it gives them immediate feedback when they make mistakes, and this ensures mastery.
Although I am impressed by the entire IXL curriculum, we only use it for math at this point. The other major component for us is Classical Conversations, which is a community-based classical Christian program. I think the CC experience will vary considerably depending on the quality of the local community, but ours is great.
Very interesting. Thank you for taking the time to respond!
I think I would still trust data over anecdotes, regardless of what Jeff Bezos says. I come across WAY too many anecdotes from unreliable people on the internet (often anti-vaccine or COVID denialists).
I think you have to be looking at well-selected anecdotes whose provenance you understand. Anecdotes of random people on the internet are actually just really bad data.
I wouldn’t accept what Bezos says very literally, but there’s a clear phenomenon he’s talking about that is probably easiest to see in the context of “evidence-based medicine”. That particular movement says that basically only randomized controlled trials count as evidence. This meant that things like aspirin, whose efficacy was established before there were such studies, needed new studies in order to “count”.
As far as I know, there is no “evidence” in that sense that a sustained yoga practice “works” for anything, but anecdotally, a large number of people seem to find a benefit for their yoga practice. The big issue is that it’s hard to even precisely formulate, let alone measure, the benefits that people identify from their yoga practice, and then on top of that there are all kinds of statistical confounds (it’s hard to randomly assign people to a yoga practice and have it stick, and there are antecedent differences between the people who do and don’t stick with it, and there’s no way to blind the assignment to the subjects, and many of the benefits take years to emerge).
This is the sort of case, where there’s a well established practice, including serious people, where I wouldn’t accept trust the anecdote (particularly from my own experience but also those I know well) over the data.
It's true that you can't blind the subjects of a yoga study, but I think you can look at objective measurements for which the subject's awareness would be less relevant. Without that, I would still be skeptical.
Sweet Jesus, this article is almost 18 thousand words long.
Scott, let me strongly suggest that you put some sort of upper bound on the lengths of these reviews in the future. Three thousand or five thousand words should be ample for making a point or three about virtually anything. The authors can always link to more elaborate supporting material if they really want to go in depth.
I liked this article a lot, so really appreciated the length--I wouldn't have liked to miss out on any part of it.
I think it's best to leave the length up to the author. Reviewers can penalize they think were unnecessarily long.
I strongly disagree, and as a parent with kids about to enter school, I found the entire thing interesting.
It’s easy to summarize long posts with AI, but hard to find detailed and well-thought-out posts in the first place.
"Brevity is the soul of wit,"
A phrase that around here is not a hit,
"Brevity is a great charm of eloquence,"
A sentiment scoffed and guffawed at hence.
Readers can always stop reading if they find the material isn't interesting enough to hold their attention. A word count limit would've eliminated Lars' excellent series on Georgism, among many other important posts.
The trouble is that if I don't finish one of the contest entries, I feel like I can't fairly vote on any of them.
Also if I don't finish a review I don't feel like I can engage in the discussion because it's inevitably "duh that thing was addressed at word 12,932".
This review started out very interesting but I lost interest in the "Does It Work?" section, whose content I could predict before I started:
1. It certainly seems to!
2. Look at these graphs!
3. But watch out for these confounders!
4. So basically we don't know, but also yes.
You are not obligated to read every single review. That's explicit: Scott asked people to pick as randomly as possible, knowing that most people wouldn't read every one. If it's acceptable to rate reviews without having read every single one, then it's okay to stop reading a review which isn't holding your attention and penalize its rating accordingly.
"Not interesting enough to keep me reading all the way through" is a valid critique, though if you find yourself skimming or skipping frequently, you might want to get your content elsewhere.
You are correct that it's not good practice to engage in discussion before reading the entire review, but if it wasn't interesting enough to read all the way through, why do you want to waste your time discussing it?
I can relate to your reaction, but I don't think that's a popular position. Judging by their votes, ACX readers consistently prefer longer essays. The book review contest winner in 2023 was extremely long too. This year 5 of the 10 longest essays became finalists, including the three longest ones. Unless Scott sees some kind of utility in restricting the length of entries, I doubt that is going to happen.
ACX readers have the highest essay length stamina of any internet audience on the planet
My friend asked me to look over a text conversation he'd had, to get my read on the situation. He warned me in advance that it was "a lot" and that I didn't need to feel obligated to read the whole thing.
I said "I read Scott Alexander for FUN. Your text conversations do not scare me".
Granted, I've also read War and Peace a few times, but only because it's the perfect book for reading before bed: interesting enough that I won't not read it, but not interesting enough to keep me from falling asleep. Also, it's so long that it's practically a new book on every read-through.
You make me want to read War and Peace for a second time. Do you have a particular translation you recommend?
The only version I've read was translated by Louise and Aylmer Maude. I enjoyed it immensely.
This is the blog with posts titled “much more than you wanted to know” (and it never is!)
Excellent piece, written ACX-style and good chances to win. I felt it could do without the complaints about their first DEI private school (but fun to read). And I wonder why you guys pays those sort of money - instead of getting a private tutor from idk Russia. Or unschool, put the school+college money in an ETF and on "graduating" your kids have enough dough for life. Numquam parum est quod satis est.
I can’t speak to author but institutional school has advantages, and a gifted school in particular has a major cohort effect. That’s part of the appeal for me, anyway.
Cursive? So I'll never get past 4th Grade-yes-I-know-not-literally.
I haven't willingly used cursive since the end of Grade School when the expectation changed to typing. Blue-Book essay exams were my nemesis.
Lol, this stuck out to me too! I was nodding along the whole time like "ah I would have loved this school so much as a kid", until I saw that and immediately switched to, "I would've flamed out and begged my parents to transfer me". I couldn't force myself to learn cursive back when there was some semi-coherent reason for it, but imagining trying to do it *now* with the full knowledge that it's entirely useless seems like it would've been even more impossible.
I was thinking at first that maybe they only need to complete a few of them, but the review seems to indicate they need to complete every single item on that list, which also seems a bit odd and against the personalized ethos.
Having seen it in practice at GT School, I believe that (a) cursive is a standin for “handwriting practice” and (b) the dosages of cursive instruction truly were not very high. None of the students seemed especially tortured by it (these were 6 to 8 year olds).
Your description of private school sounded so awful it made me nauseous. I can't believe people even consider putting up with that for 35k a year
This was a good review!
This might be a dumb question. But isn't homeschooling pretty similar to 1:1 tutoring? If the parent doing it is educated, motivated, and following something like The Lost Tools of Learning substack (or something else, I just mean a well-rounded curriculum), I would think that at least in the elementary years it would be pretty similar. Sure, it might not be exactly 1:1 (as I'd expect most homeschooling parents to have multiple children), and the parent is probably not going to be a subject-matter expert in every topic the kids ask questions about, but teaching them to research and find the answers is also valuable. You've still got rapid feedback and tailoring to the kids' needs and interests.
Why, then, aren't all homeschoolers Polgars, and why do the homeschoolers in Alpha do worse?
Hypothesis: the *typical* homeschooling parent does not put in anything like professional full-time attention to their child’s academic motivation or the wider motivational system in which academics occurs. Obviously there are going to be a lot of exceptions on the right tail, which is why you get Polgars and why some of the homeschoolers in Alpha do well, but it would explain the modal outcome.
I think a big aspect is also selection bias. I home ed because my son is autistic and I would say well more than half of the homeschoolers I know are doing so due to some form of disability; in some cases, cognitive disability.
The other issue for more religious families are large families. It's extremely hard to home school if you are also taking care of under 5s. And there's also some evidence they under-educate girls purposefully. There's also the unfortunate trend of classical education which leaves children with significant content gaps.
We also know the recent trend of 'unschooling' tends to result in pretty poor performance.
If think if you took a subset of families with a secular, academic educational philosophy, who were neurotypical, and there were no more than two kids in the family, the results would be better.
I suspect this actually is most of our homeschooling families but I haven’t dug into the data. Our primary homeschooling beta testing audience at the moment is people who say “I want to try this but cannot move to Austin”.
Points taken about the range of homeschooling contexts though.
You're definitely right in general. But also my colleague Matt Bateman is correct that the latter description fits those piloting the Alpha homeschooling program thus far.
If these smallish family & low special needs users have trouble achieving good results without the in-school motivational model offered by bricks and mortar Alpha, a general (homeschooling) audience would likely find it even harder.
Maybe parents are more susceptible to accommodating their child's natural inclination to be a bit lazier than they would optimally be. "I've done so much math today, can't I go do [other thing that is also constructive but is not math]?" is easy for a teacher to say no to, but maybe not Dad.
The academic results seem impressive, but is there any data about the effects of that amount of screen time upon developing brains, or are such concerns only made up by neo-Luddites? Or is it that screens are now ubiquitous and therefore any attempts to reduce screen time are futile? Just suck it up and accept that human brain development and eye development are potentially permanently affected? Assuming that the trade-off is that accelerated learning will mean more time for other potentially non-screen pursuits, we just might have to accept it the way we have to accept the eternal presence of microplastics in our blood and tissues.
There is no research indicating that 2 daily hours of screen time (which is not very high, and not dissimilar from typical in the majority of American public & private schools) is detrimental to elementary+ age kids.
Indications that kids under 5 should be limited to no more than one hour of screen time daily. After that age, inconclusive - some evidence pointing to higher rates of depression & anxiety in teens as daily screen time increases, but these studies (at least ones I've seen) don't typically differentiate between academic screen time and social media/gaming use. To me there's obviously a major qualitative difference between two hours of reading/science/math practice and spending your entire evening scrolling TikTok and Discord.
Thanks for the information. I appreciate it, and agree with you about the qualitative difference between one type of screen time and another.
I think most complaints about “screen time” are really about the junk that it’s filled with, whether it’s algorithmically-curated social media or short format video, or trashy freemium games. If you could ensure that people were having live video conversations with interesting people, or watching classic movies and reading classic novels and playing deep video games, then the screen time itself doesn’t seem like it would be as big a problem (especially if there’s something making them get up and do something outside every 20-30 minutes so their eyes refocus).
Are there screen time studies that adjust for content? And speaking of breaks to let their eyes refocus(which I agree with), are there any studies that specifically look at attention span differences with regard to amount and/or content of screen time? i.e. long-form video essays vs. "shorts"?
Am I reading this right? Now, they’re collecting 40K in tuition from extremely bright children with motivated parents and teaching them in a specialized environment full of enriching externals but running at a loss. The author says the software piece is fine but basically useless without that structure. And their financial plan is to massively expand — to reach children who are not naturally bright, and do not have motivated parents, and to strip away the non-essentials, which the author says is necessary to make the program work. Doesn’t this sound like the classic tech disruption strategy, but for education? Offer a very good but financially unviable product to start, drive the competition out of business, and then offer the true, shitty, profitable product once there are no longer viable alternatives?
There are many bright children at Alpha but we don’t do an academic screen. Lots of students come in behind.
The “massive expansion” is stepwise and will occur across models with differing cost constraints and price points. I’m sure we’ll learn a lot and have to adjust a lot. The unit economics are not at all crazy. A lot of the R&D cost will *definitely* go down. The AI model costs right now outstrip almost everything else.
They only screen at GT School which is the gifted program, but I believe they don’t screen kindergarteners, only kids with previous schooling.
Great review!
I'm not American so forgive me if this is a stupid question - it sounds like Alpha children complete the full curriculum in half the number of years the state expects? What do they do when they've finished the curriculum? In the UK there basically isn't any way to skip from primary to secondary school early, so if this model was expanded to the UK it seems like there would be a few years a thumb-twiddling built in. But presumably the solution to this problem is so obvious to Americans that I don't think I see a mention of it in the review.
Presumably they just start taking university-level classes? This is how it generally works in public school as well to a limited degree (e.g. AP classes).
While it is also very difficult for most bright children to 'skip ahead' in America, it *is* possible, and the main roadblocks are typically actually at the institutional rather than state level (because most institutions do class-based instruction grouped by age & ability, they do not want kids getting more than a year ahead because then the kids will be with much older children who are not at the same level of social and emotional development).
After that the hard part is basically just finding a university willing to accept a young student (though this does happen, so clearly it's not impossible) or alternatively building a post-curriculum after the state curriculum is completed.
(I am parent of Alpha students who's gone to work for them) Not stupid! - basically high schoolers begin to move in different directions. Some students are set on entering very competitive colleges, for them doing the minimum to have finished high school equivalent is not enough, they continue to do math/literacy to prepare for SAT (college entrance exam) and Advanced Placement coursework (college equivalency program). Others move more in an entrepreneurial direction, for instance.
You can see more about the AlphaX projects as they're called here https://alphahigh.school/alphax/
The gifted and talented version of the school essentially aims to have students college ready academically by the end of 8th grade though it's new enough that I don't exactly know what the first classes of graduates (several years away) will in fact choose to do. Some probably will enter college before age 18.
You can actually go to college a couple years early in the US without any significant difficulty, and earlier than that with some more effort and arrangement with the school.
For anonymity reasons you should probably blur out the name in certain screenshots; anyone familiar with a certain educational Substack would recognize it.
I didn't read the whole thing, because it is way too long and not particularly interesting to me personally. I stopped about halfway through and skimmed the rest. However, from what I gathered, this seems like a great program for gifted kids, and not so great for other types of kids, especially ones that would struggle with the pace. But then again, gifted kids will do great in just about any type of program, especially if they are self-motivated. So what's the point in spending $40,000?
I think the point is Alpha is self-paced so they go faster with less time wasted on bureaucratic busywork. It leverages AI in a way that semi-simulates a 1-on-1 tutor experience. It's an improvement over traditional private schools which people have already demonstrated a willingness to pay 40k a year for.
It seems promising but my guess is the real secret sauce is selection effects and enthusiastic founders. Wealthy early-adopters are a highly-selected cohort, as are educators willing to take a chance on a radical new teaching model. That's unlikely to do well at public-school scale, particularly in misbehavior-prone urban districts. But it might be a viable alternative to both magnet schools and traditional private schools.
The quote is deeply ironic in that the point of teaching the child is for the child to learn, which using AI undermines. The point of the teacher teaching is for the child to learn, which using AI for that half does not undermine.
There may be other reasons but not that one.
Surprised Austin private elementary schools are so cheap. Elite Atlanta privates cost $35K+ for elementary school now.
"20:1 ~teacher~ guide:student ratios (vs the 5:1 ratio at the Alpha private schools)"
Are all of these teacher:student ratios inverted throughout the piece? It's very confusing.
Yes they are and it’s surprising to me that no one else pointed it out, it was incredibly confusing
I thought the 5:1 teacher to student ratio was pretty crazy, then he mentioned public schools having a 25:1 ratio and I realized what went wrong. I could see some school for rich kids that has five people teaching each student, but I've certainly never been to a school with 25 times as many teachers as students.
To clarify, there are five students to one guide at Alpha.
As homeschoolers we do some of the same things; we use anki (which is free!) for spaced repetition, we do a sustainable amount of academic work daily rather than pausing for the summer, and of course we don't have to try to educate thirty kids at a time. These make learning vastly more efficient and Alpha School didn't invent them. But we also do everything we can to surround our kids with adults who are enthusiastic and open about their own lifelong learning (ourselves included) and with materials that are worth their time (a thing I wouldn't say about IXL).
Humans are very driven to learn, care what the people around us are doing, and are also mostly at least a bit demand avoidant, so it has always seemed to me that homeschooling parents can choose between spending their time explaining why kids have to do math but adults don't, doing some tricky alcumus problems themselves, or paying their kids and hoping the kids don't find a better revenue stream.
I have no issue with paying kids to do things that they are free to refuse to do, but as a kid I would have found it insulting to be required to do something and then paid just a bit for doing it. I understand that other kids are more pragmatic though, and are happy with the arrangement. I also would hesitate to pay a kid to read a book because there is no clear definition of reading a book well enough, and I'd rather read some of the same books they do and enjoy talking about them together than quiz them and incentivise the opposite of close reading while becoming the did-you-actually-read-it-though inspector.
This is doubtless racist, sexist and homophobic, but Alpha seems to have a lot in common with schooling methods during lockdown. And Bastet knows that those methods were disastrous for educational outcomes.
The one exception I know of personally was one ant colony, and that was simply because the queen ant took homeschooling to insane levels. Before, the little ants had been begging to be homeschooled. Soon, they were begging to be able to go back to school.
Which brings me to my real point - for decades now, humans have been pushing various nostrums to address the fact of Why Can't Johnny Read? None of these nostrums and magic spells perform as advertised, although many of them do enable educational consultants to make bank.
Fact is, parental involvement, parents that ride little Johnny like they are trying to win the Preakness, combined with an environment that stresses learning, is the only thing that works. Nobody wants to hear that, because other than that queen ant, nobody wants to put in the work. Nobody wants to point the finger at the parents. Far easier to look for magic bullets.
There are some really obvious cases of ChatGPTese sprinkled throughout this review, yet it's also clearly thoughtful and original, with lots of concrete and specific claims. Very unusual combination in my experience.
I hope once the contest is over and things can be de-anonymized, the author would be willing to comment on how they use AI in the writing process yet produce a high-quality non-slop document. Although I do worry the answer is just "you have to be literate, and also actually care".
ChatGPT has a writing assistant mode. You can tell it what your central idea is and have it produce a rough draft. This guy probably did that and then rewrote it, leaving some of the original in.
Would be interested to see some specific examples from the text.
most obvious to me was this paragraph:
"Elite private school attendance buys you smaller classes, brighter kids, and fancier field trips – not academic acceleration ... That prior—show me velocity, not polish—is the lens through which the rest of this review should be read."
What about that says LLM to you? Is this just em-dash discrimination? They're not even consistent, which I'd expect from something copy+pasted.
The highly telegraphic writing style and punchy parenthetical phrases in italics (“show me velocity, not polish”) are in my experience extremely characteristic of o3.
Excellent article. I'm a bit more skeptical than you are on the magic of Alpha School compared to other schools, academically, and its ability to scale to a true mass market product. However, the tech-heavy model does seem like a way to cut expenses ($40k flat) compared to Horace Mann ($64,070 + a lot of extras + an endowment) or Harvard Westlake ($49,700 + a lot of extras + an endowment) for an elite school, and even if they have to drop some of the international travel for grade schoolers, it might be a sustainable long-term business without billionaire financial support. I hope it's still available when I have kindergarteners.
$40k/yr for 17 years to have a 95th percentile achieving kid is an incredibly good investment.
hmm education is so complex
Quick story about another student who goes by Austin Scholar here on Substack.
When I first met her, she was a sophomore in high school (15 years old). I was teaching writing at the time and actually rejected her from my program because it didn’t seem right for a teenager to participate in a program designed for high-powered adults. To her credit, she pushed me to get in the program and after a short interview, I reluctantly let her in.
There were ~250 students in the program and she was a top 10th percentile writer as a high schooler. Watched her get a perfect 1600 on the SAT and get admitted into Stanford. Even as I’m writing this, she’s doing an internship with me for the summer and one reason why I’ve been eager to hire Alpha students for various projects is that they’re uniquely self-reliant because of the lack of teachers in school while also being uniquely on top of things because of the strict standards that Alpha maintains.
All this is to say that Alpha doesn’t work for every students (and it’s not designed with that in mind), but when it works, it works pretty darn well.
She’s written about her time at Alpha if you’re curious to hear a student’s perspective: http://austinscholar.substack.com
As someone who has done casual inference professionally for a very long time, I can say with confidence that Caplan's work is an exemplar of Gelman's many-forked path, where the results were mostly determined beforehand and confirmed by choosing the right path through the data. Not work to be taken very seriously.
Can you elaborate on this?
I don't have time to do a deep dive. This critique of one of his books gives you some idea: https://www.educationnext.org/main-purpose-of-schooling-review-case-against-education-book-caplan/
Basically the things he's trying to claim are very hard to identify in the statistical sense. And extrapolating even from studies where ~everyone would agree the causal inference is solid is always low confidence because of shifting context and the specifics of the outcome. This means that it's a matter of judgement for every analysis about what the results are actually telling you. This would make the kind of sweeping statements he makes suspect no matter what. (The deep dive into a specific analysis of his is what I don't have time for).
But there are a number of other clues that you shouldn't believe his judgement calls aren't always strongly biased. First, his shocking lack of technical skill. His anti education book was based on 15 months of building an opaque tangle of spreadsheets. Which he then complained no one double checked and reanalyzed 🤦♂️ This alone is near disqualifying. It's almost certain he's doing mediocre-at-best casual inference when he's limited to what's available in excel. It also makes his work opaque and pragmatically unreproduceable.
Second, he writes a mixture of opinion/ideology, data reporting, and results from his own spreadsheets. He will often try to pass off his ideological commitments as facts. Essentially I'd argue that he's revealed his commitments. Everything he does is work to bolster his ideological commitments through data mining and argumentation. He spends a lot of time on polemic and constructing logic chains that he thinks make his analysis solid. But he's never spent the time gaining the technical skills to actually do best-in-class non-experimental casual analysis. In the same way you should be skeptical of left wing academics whose work always just so happens to "prove" that whatever flavor of leftist campus activism is hot right now is true, this shouldn't give you a lot of confidence in all the methodological choices he makes in constructing his data argument.
Alpha's marketing is excellent - and there are many educators and parents I admire who are excited about them. I've been craving more detail and this answered many of my questions.
On the pedagogy of app-based learning & measuring using MAP:
This approach seems solid for the basics of math and reading, though this piece's point that reading and discussing longer form books is missing is key. Much has been written about the negative impacts of over-optimizing for short passage reading on the literacy skills of American young people. The short-passage only approach is usually associated with public and charter schools, while the Harvard Westlakes of the world go all-in on literature.
Science is even trickier to teach via apps and measure via standardized tests. Many schools don't take MAP Science at all, as a note. IXL works well for learning baseline science content, but I have a hard time imagining multiple-choice modules go far enough to produce mastery of science skills as written in the Next Generation Science Standards. You can read those on their website, but they are largely complex and applied skills such as 'developing models,' 'representing and graphing data,' and 'making evidence-driven arguments'. IXL teaches these skills by having learners 'fill in the blank' of a partially completed graph or argument. The MAP Science test does the same, because it's also a multiple-choice measure. Isn't this fill-in-the-blank mode of thinking much easier than asking a learner to actually demonstrate the skill (figure out how to represent and graph data) themselves? If I were an Alpha parent (especially one not at the GT school with the extra academic competitions, which sound great), I would worry the science learning isn't rich & applied enough.
Not sure how History/Social Studies learning is assessed at Alpha, since MAP doesn't have a Social Studies test. I assume just in-app and their mastery tests. Similarly to science, I would worry IXL + a sprinkling of reading passages in AlphaReads lacks the richness and depth I'd want. To learn Social Studies well, I'd want them to follow a more comprehensive trajectory that includes longer sources, group discussions, research projects, etc. I'd worry that the afternoon activities as described are too piecemeal to succeed in this area but could be wrong.
Not surprised Writing is a miss in this parent's eyes - I've never been impressed with any edtech writing app. There are a few that are nice for drilling grammar and can supplement writer's workshop style teacher-led classes.
These things feel surmountable. I wonder if Alpha has considered using a portion of afternoon time for teacher-led, Hoel-style 'aristocratic tutoring' in humanities and/or science to see how well these in-app learning gains transfer to real-world applications and open-ended conversation, while remaining hyper-personalized.
Also, I was surprised to see IXL as the app of choice. The questions in IXL are quality, but it's boring as hell - the forms of engagement are unvaried and dry, and it must feel repetitive for kids to learn in the same mode every day for up to 12 years. Last time I spoke to somebody at Alpha, they said they used several apps for each subject for variety and to ensure coverage - perhaps this has changed, or maybe just wasn't mentioned here. They'll need to find a way to make the incentive system foolproof for parents to implement in the homeschool/virtual model.
Early in building Prisma (virtual program I work for), we experimented with an app-driven morning schedule. We ended up retiring that approach for every subject except math, for some of the reasons above. This year, our kids grew 2.3x as fast as their peers nationwide in math, and 1.6x as fast in reading on MAP. We definitely could read this as evidence that apps are the best method and should be applied to the other subjects. But our parents tend to choose us for reasons other than optimizing these scores, and would hate if we abandoned our discussion and project-based Literacy curriculum for IXL. Could we supplement with apps and boost those further? Probably! Overall I'm excited Alpha exists. They're developing some cool innovations that can meaningfully improve the space.
This article was a great read, and I agree as a parent of GT School (Alpha’s school for gifted children in Georgetown). My son is 10 and taking 6th and 7th grade classes. He met and/or exceeded all of his end of year MAP goals and continues to be in the 96-99th percentile across the board (one reason we entered GT School to begin with). He also has developed a true love of school. To the point where he sometimes complains of we take him out for a day for a long weekend. He especially loves his guides, who routinely find ways to customize rewards and incentives to each child’s interests. My son won a Lego chess set and Mario stickers amongst other things with his GT bucks. Their reward system is highly effective and the personalization acts to make students feel seen and understood. Though the author is correct that I don’t think this model will work for every child, my son has ADHD and is actually doing spectacularly well even without his pervious IEP accommodations. Their low ratio and individualized learning model have been both impressive and effective. This all coming from someone who was highly uncertain and critical of the idea of AI teaching my child.
The website has a much weaker 100% money back guarantee than the one stated here. Can anyone find a link to it? Perhaps they've walked it back as guaranteeing every child wins a national award seems like a good way to lose your shirt.
It is only the GT school. It is still there at the bottom of the homepage: https://gt.school/
I see people are interested in hearing from the parents — I am one, and I'm NOT employed by the company. My 7-year-old is at the GT School (which, btw, is $15K/year; Alpha's specialized academies have lower price points). AMA.
To start, here is a bullet-point list of why we drive 30 minutes to the school (TL;DR: it's worth it. We will move closer now that we know it works for our son):
- Compressed academics utilizing the Pomodoro technique — work intervals followed by breaks; much more effective for high-energy kiddos who can't yet focus for extended periods of time
- Self-pacing — go as fast as you can or as slow as you need in each subject; no shaming through grades, no "B student" stigma
- Verification and accountability — tests are used in a productive way. Apart from three-times-a-year MAP testing, there are mastery tests whenever the student completes the lesson plan for each grade. I wasn't a big fan of the testing part going in, but they use the detailed data from the test report to identify gaps and then serve additional supporting lessons in those areas. My son completed 2.5 grade levels in one year, but I'm more wowed by the fact that he actually mastered that material than the speed itself.
- The guides!!! — well-paid, highly motivated individuals who are not burned out from a mountain of paperwork and can truly engage with the kids
- The group workshops and individual pursuits (check chart) — the element of real-life skills, learning how to bring a project from concept to fruition, on your own or with a team — this part was really important to us
- Growth mindset environment — "the power of yet," philosophy of high expectations with high support, finding your "why"
- Community — engaged parents who are willing to take reasonable risks to try something different for their kids
It's not the AI, it's not a magic bullet — it's not one thing; it's the whole package.
And before you say, "Your kid is already gifted, so he would do well anywhere," that is a common misconception. Gifted does not equal high-achieving. Our boy is extremely high energy, motivated mostly by his interests, with spotty attention-paying ability. We are quite certain he would not thrive in a conventional schooling setting, and we have spent countless hours researching and considering various education models. This is not perfect; it's just the best we found.
Also, just for fun — and also to address the "wealthy kids" stereotype — we fall into the "first-generation immigrant family" pool, making sacrifices to give our son the best shot. I know every parent wants that for their kids, and I'm hoping my long comment will be somehow helpful in your journey.
I am the parent of two Alpha students, they are moving to the GT school this fall. As for me, I've been drawn out of more than a few years of stay-at-home motherhood in order to get involved in the Alpha mission.
We initially chose Alpha just because all the other schools available schools seemed blah. My husband, Byrne Hobart (an independent finance and tech writer) is one of the non-conformist types mentioned in the piece.
But my older daughter turned out to be exceptionally bright and her results at Alpha have been way beyond expectations. With no homework at all and very little drama, she has maintained not only 99% achievement but also top several percentiles of growth for 3 entire school years.
I never heard her complain that she was bored, never had to meet with a teacher to grovel or battle for acceleration. At home we focus on reading for pleasure, arts and crafts, and exploring niche interests (she's obsessed with penguins).
Intense acceleration offered in a pleasant environment has been a priceless gift to our family. I never worried about oldest's education while I was pregnant again and caring for an infant.
If it means anything to you, this is what o3 says when offered my daughter's MAP scores: "[X] is operating at an 11th-grade academic level while still in Grade 3 and is continuing to out-grow almost all comparison groups. Alpha.school’s 2-Hour Learning model is not only keeping pace with, but in several dimensions outperforming, America’s most elite K-12 institutions—all while freeing up 60-70 % of the day for deeper projects and community-building, exactly in line with the school’s founding mission."
My second daughter joined Alpha a year ago and she's a tougher cookie (2e/on the spectrum). Even with far less than 2 hours per day on the apps (despite her wonderful guide's ceaseless efforts and encouragement), her growth has still been immense: https://x.com/gtmom/status/1931058079745056974
Very happy to answer any other questions, here or privately.
meh, nothing about this is particularly different from other educational fads, because they all share the same commonalities:
1. cherry pick the most gifted and motivated students
2. target parents who are both rich and overinvested in their kids education.
the thing is, those two things can make any kind of system work. I see this with homeschooling; everyone glazes it the fuck up, but its horrific otherwise if the kid or parent just can't throw heroic levels of energy at it. That's when teachers help, not systems per se, because its very easy to game a system. Hence chatgpt writing papers and passing around test answers. or cheesing curriculum requirements.
and honestly you all overrate intelligence and underrate connections and money in life success, that is why your kid does well or not short of mental illness or parental tragedies. Education is probably not going to help you beyond a point, and there is some truth to Caplan.
(only at the extreme levels, though: its annoying he is so stupid and doesn't realize he needs a widely educated base of people to keep his university running, and that tenure hothouse people give far less value to life than he things)
#2 is fairer than #1. Alpha admits plenty of students who are behind academically and not especially excited by school. GT School screens for academic potential, Alpha absolutely does not.
Re: cheesing curriculum requirements, the students sometimes do try to do this, and it gets patched. It’s a fun arms race.
I don’t think the premise that “any system works with rich kids” is particularly true. A lot of very specific works goes into helping these students, and I don’t really see evidence that it’s just a wash compared to very different methods at other schools, and see plenty of evidence that the school is helping these students. (Other elite schools don’t even publish their K8 testing data.)
behind compared to who, though? parents who give kindergarteners IQ tests? other elites? i really don't think there is that strong a difference, and that's not getting into parent differences in how tiger mom they are.
the cheesing i just mean that without motivation by some one systems fail. id bet the weak point is the brazil teachers, lol at tech bros offshoring. better hope they don't fake answer calls, or exploit the system because they can.
and no those kids would be fine without that school. places like alpha bank on that, because its really hard to actually teach struggling students. little effort of maximum profit is the good way for modern business men
There are multiple comparisons of interest. I think with Alpha in particular the most helpful comparison is other elite schools, yes.
The “brazilian teachers” are software developers and academic specialists who work on the curriculum and platform. They run the coaching calls because they are the ones who need to know what’s not working and fix it on the app side. Not sure about Brazil but some of them are indeed remote.
> meh, nothing about this is particularly different from other educational fads, because they all share the same commonalities:
This seems like an oddly defeatist attitude. Everything you try will appear to work, therefore there's no point in trying to figure out how to do it better.
> and honestly you all overrate intelligence and underrate connections and money in life success
This I agree with, but it does seem like Alpha School is aware of this and trying to deal with it, with "life skills" (of the sort practiced by mid-level managers at large tech companies) being a large part of the curriculum.
they arent trying to do better, they are trying to do cheaper and get quick bucks.
like you don't need to keep reinventing a stable solution; a private school with correct teacher/student ratios and managed well is perfectly fine. But to run one is expensive and modern businesses love their mvp and getting something for nothing. They'd love it if your kids teach themselves with ai lapdogs and the occasional overseas (got to pay as little as possible!) check in.
honestly businesses are not looking out for your own good.
the life skills...eh its a skill to know the right people i guess, my point is more the kid being smart or not matters less than the dad being famous, which zip code you live in, if dad overloads you with extracurriculars, or if parents can bankroll first business or internship. or failure might be from things education cant help much with.
think the chinese dudes know with guanxi, but we kind of love our horatio alger.
> like you don't need to keep reinventing a stable solution; a private school with correct teacher/student ratios and managed well is perfectly fine
It's perfectly fine, but is it optimal? My K-12 education was perfectly fine, but could have been better.
> the life skills...eh its a skill to know the right people i guess, my point is more the kid being smart or not matters less than the dad being famous, which zip code you live in, if dad overloads you with extracurriculars, or if parents can bankroll first business or internship.
Right, that's exactly the sort of life skill I wish I'd learned at school and didn't. I emerged from school with too much faith in the idea that just "being smart" is the route to success, and ignoring all the other aspects.
I, too, felt some instinctive revulsion to the idea of incentivizing children to learn, and I think I know why it feels wrong: it might actually work.
No matter how you dress it up, school must involve adults compelling children to behave in certain ways. People who like freedom find this somewhat unpleasant to think about, but most agree it's for the greater good. But incentives are just a much more effective means of compulsion than other methods. At least when a teacher hits a student it's clear what's going on. Incentives are, well, seductive.
Reading this essay makes the whole school feel very creepy. Points for compliance. Eye-tracking software. When we're worried about socialization, we set up a new system to reward that too. The kids in this system don't really seem to have free will -- they are allowed to make choices, but only within the system.
I don't think this is necessarily a bad thing! If we're going to control how children think and behave to educate them, we might as well do it effectively. This school sounds a lot more fun to be in than the one I grew up in. But it is very much the Brave New World to the public schools' Nineteen Eighty-Four, and it's hardly a surprise that a lot of people find that scary.
Fascinating read. Really appreciate the context outside the school as well as the deep dive into the entire methodology.
There is one big, conformity forced assumption here and that's competition. Absolutely everything in western society is a competition and we've been taught that is the best way and the only way to live. Our kids are measured by percentiles and only if you are the best of the best can you ever hope of being accepted into the best colleges and once you have the name of that institution on your resume then you can work for the best firm to make the most money and then you'll be happy. The problem of education is always framed as "how can I make sure my child WINS?"
For all the talk about nature vs nurture and Caplan's work, this entire post reads like signaling. “Insofar as schooling makes you healthier by raising your status, its health benefits are zero-sum: you can’t raise your rank without dragging others down” (Case Against Education, p.171). I don't believe your claim that after love and respect the best thing you can give your kids is time. Or rather I believe the claim, but I don't think you believe it. Why move your family across the country in the middle of the school year and spend $120k a year on that schooling? You didn't do it to give your kids time. You did it so they can make the same kind of money.
How do I know it isn't for time? They don't even have a free half-day. Chess competitions, quiz competitions, essay competitions, property management (I'm so curious to see how a bunch of 5th graders would decorate an Airbnb). Even if you went to the sports academy it's the same game. Parents will spend hundreds of thousands of dollars on sports training to beat other kids. Beat all the other kids and they can go to those same colleges to get those same jobs. A huge percentage of Olympians in the 2024 games from around the globe were currently attending top American Universities. In that context as you mentioned over-optimization runs the show. And if they finish school before college age are you telling me they will actually spend 7 years of their life with free time to do what they love? Maybe some will, but the kind of parents enrolling in this school are likely going to push their kids to finish college early and make money earlier.
This kind of schooling may require a certain kind of non-conformist to enroll, but you are still conforming. I'm certainly not telling you how to parent and I'm not saying money doesn't matter. I'm saying is I don't experience any happiness from "keeping up with the Joneses".
At GT School, competition is opt-in. The students find it fun. It’s a culture of *friendly* competition. Really remarkable—that culture is actually the thing that sold me on the school.
I completely agree. My son enjoys the competitive aspect, but also excitedly tells me about his friends beating him. GT School is comprised of high achieving kids that like competition but the guides are very gifted at making everything friendly and cooperative at the same time.
One of the main reasons we're sending our kid to this school is because this model is not based on a zero-sum game like - I would argue- most conventional schools are. I can't comment on the author's ambitions for his kids but ours are simply that he finds an occupation that brings him joy and fulfillment and finds a way to support himself and his family doing so. Many ways to skin the cat, many ways to live a happy life and this model offers more time and freedom to explore those ways, while still maintaining accountability for the basic core skills, than others that we looked at.
Fair enough. I'd probably send my kids to this school too simply because I'd want them to learn as much about the world as they can. It seems like they really excel at that. And no matter my objections it's not like other options are going to be so drastically different for the end goal. We're all just doing the best we can with what we can.
My frustration is more about the problem we are trying to solve here. The post is focused on academic measurements, percentiles, and the marketing material feels aimed at winning the zero-sum game. Learning itself isn't. But as the author points out the natural next questions are about getting into a good college and getting a good job. The questions almost all parents inevitably ask themselves.
I recently talked to my cousin who went through an accelerated program, started college 2 years ahead. I asked what she loves about her chosen field/major, basically what drives her in school. She said she doesn't like it, but it'll make good money. Her parents and every single adult in her life talks the same way. It's always about making as much money as possible as early as possible.
It may be impossible to not play the game no matter how pure your intensions. It's just so culturally ingrained. Personally I'd want us to solve the problem of "loving to learn". In many respects I imagine this school does a great job of that. However I often think of Goodhart's Law: When a measure becomes a target, it ceases to be a good measure. We are measuring academic performance relative to other kids. Our hyper fixation on comparison in academics or on social media can really suck the joy out of living if you let it.
I loved this essay but your comment here captures most where I land.
We humans with fairly finite lives can only serve one or two priorities at a time. If getting ahead, faster, so that one can "win" the money game is the priority, it will crowd out quite a few other good potential priorities.
In the video of the little kids each talking to their own individual AI about how to do their hobbies, one of the boys asked the AI, "And what do I do now?" The AI was instructing him step-by-step what to do to for his arts and crafts. *To what end?* What happens when we can't really guide ourselves any longer?
I really liked this essay and would also love to send my kids here if it was cheaper. But I agree there is something deeply unsettling.
I also read thezvi’s childhood roundups, and while I like those they make me bristle at times about the absolutely relentless min-maxing. I have been considering posting a comment similar to yours on Zvi blog actually.
I feel like the optimization bug has infected me and sucked the joy out of my life. I suspect it became ingrained over a million hands of online poker, or maybe it’s an OCD thing. I have to justify everything in terms of EV and I hate it. I don’t want to pass this on to my kid.
Very impressive essay.
I was caught off guard when the author mentioned using Duolingo to teach foreign languages. I thought it was fairly well established that Duolingo isn’t particularly effective for learning new languages, due to the lack of active recall, among other factors.
Yes. Take that insight about Duolingo, and consider that the other educational apps mentioned are similarly engineered. I wrote a lengthy comment elsewhere in this thread elaborating, but this is a real concern - does bite-sized, multiple-choice module learning transfer meaningfully to an authentic context as well as it does a similarly designed standardized test?
There's evidence that it may - SAT scores are the best predictors of college achievement, for example. But I wonder if Alpha students' test scores represent their real-world intellectual ability as accurately as a kid who spends much less time on 'test prep' style learning.
Maybe Alpha's mastery exams are designed in such a way to mitigate this, too, I don't know.
We agree that Duolingo doesn’t work. The students wanted to try it last year at GT School for various reasons, but it’s not part of the platform.
Re: transfer, overfitting to tests, and achieving deep understanding, I do think that’s a risk area for us. We have a different pedagogical risk profile than most other alternative schools. I think we’re managing it well but it’s ongoing. I definitely don’t think it *trades off against* the methodologies we use, but we have to be sure we’re achieving it.
It was made particularly worse by the fact he complained about the effectiveness of language classes in regular high schools with human teachers. And he brings up Duolingo as a better alternative????
I agree. Duolingo is useful for vocabulary memorisation, (very) basic reading comprehension, and habituating the learner to a range of native-speaker accents. But its commitment to bite-sized chunks of content prevents it from teaching grammar in the clear and explicit way that's necessary for language learning. There are many grammatical concepts that simply cannot be taught from scratch in five minutes flat.
This is one of the best things I have read all year. But I can’t help being just a bit upset.
As someone not involved in education, it has always seemed to me that an approach similar to this is, as kids used to say, “a no brainer”. If I had a billion dollars and wanted to design a school to optimize learning, it would look something like this. This is what I remember worked best when I was in school, and new technology would just empower it more each year.
The reason I am a bit upset is because I have heard for decades how “nothing works or scales in education”.
I am starting to think that the problem in education isn’t the children, it is the dysfunctional mindsets and frameworks of most people involved currently in education.
Obviously this won’t work for every kid, and is currently not affordable for most, but I think something like this could soon make a difference for millions of families.
"Some kids will rush into college classes. Some will choose to use the time to play sports. Some will use the time to master chess or quiz bowl or programming. Some will take time to travel the world with their families."
If we go by the South Korean case study, ~all kids (or all kids of ambitious parents) will choose to spend that extra time on more and more classes, as competition for "elite" colleges and the "elite" jobs they entail only gets tougher and tougher. In the end this just increasing the price of signalling even further, leading up to a very expensive equilibrium.
Not all kids! My son enjoys the 2 hour learning model because he is bright and that’s all he needs, but also because he finds school work boring compared to workshops. He aspires to be a gaming programmer and we won’t pressure him to attend a fancy school to do so. Wealth doesn’t always equate to elitism. I’ve found that most private school parents just want happy, thriving kids, just like any other parent.
That sounds wonderful and I don't doubt for a second that this is a great environment for a kid right now. In fact, if I was living in Austin and had that kind of money, I'd also consider this school for my son - mostly for the wide plethora of cool and exciting things the kids seem to be doing in the afternoon according to the review.
What I am skeptical about is how such a system will evolve at larger scales (if it is as effective as advertised). Consider this scenario:
You start with Alpha school with 2h learning in the morning and workshops in the afternoon. Then someone comes up with Beta school that uses the same method but ramps up the learning to 3 hours to get 40% higher test scores (let's assume some diminishing returns). Gamma school is at 4 hours, Delta at 5. The alumni of each of them crowds out the alumni of the previous one from elite institutions, providing constant selection pressure. By the time you know you have rooms full of kids glued to computer monitors for 8 hours a day doing spaced repetition pomodoros.
There is currently no market for this. Most parents, even parents shelling out for Alpha, believe that the 2x learning promised by 2HL is already possibly too much/too fast, not too little. The scattered parents who are like this satisfy their preferences with tutoring, math camps, etc added onto other private or public school educations. They do not, in the U.S. at least, form an interest group that drives whole schools into or out of the market.
Though I do wish that there were some secret cadre of academic accelerationist families ready to book out GT school and pressure us for even more right away. Would be a great problem to have, to be banging the drum that their students need life skills and not just academics, rather than explaining how little students learn at regular school over and over.
(Am Alpha parent turned Alpha employee)
Phenomenal piece. It highlights a point that any rational observer of the lock-step "product" of American education might ask: why not let kids progress at their own pace, tailored to their ability, and focus on mastery instead of simply promoting them annually by default? You're not claiming to have a gold-standard study here, but honestly, it’s not hard to imagine how well that kind of model could work.
That said... I felt a real sadness reading it. The image of kids drilling on computers—even with breaks—was disheartening. My reasons are entirely anecdotal, and I can’t back them with empirical evidence, but I went to a remarkable private school from preschool through 8th grade. My cohort in the "Honors" track was made up of extraordinarily bright, talented kids who later became academic, economic, and artistic achievers. Total selection bias, no doubt—but here's the thing: we loved school.
Why? Because our teachers loved teaching. History came alive for us every year—from the Egyptians in 4th grade to post-WWII in 8th. My lifelong love of literature was sparked by our English teachers reading aloud with passion. If you've seen Dead Poets Society, it felt like that. We wrote and wrote. We adored our science teacher most of all—my closest lifelong friend, now an M.D. Ph.D. and Chief Science Officer at a biotech firm, credits that teacher with changing his life. Another close friend, a composer, points to our Head of Middle School—a master musician who packed our days with Glee Club and Orchestra—as his inspiration.
I don’t quite see that at Alpha School. There’s no shortage of engaging extracurriculars, and maybe kids just self-select into what resonates with them. Still, I don’t see love and passion. And I get it—love and passion don’t scale. They didn’t even scale within our own school. The honors track received all that energy; the median cohort—still smart, still tuition-paying—got "good, not great."
Maybe what we had was the modern equivalent of a Victorian-era tutor: personalized attention in a small enough class where everyone could thrive.
The love and passion are there. I wouldn’t be sending my daughter there next year if it wasn’t. It’s hard to capture and convey in the same breath as one is talking about the software platform, but they are there.
I get the worry about wanting inspiring teachers and role models, though. I had this worry too, before I visited. The teachers do serve this role, it just has a different specific look than it does in Dead Poets Society.
OK, that is wonderful. It would be interesting to try and quantify the "inspiration" effect. We talk about how the data says, well, it's mostly your genes, and maybe your peers. But so there is so much anecdotally about inspirational teachers and professors, that maybe "the data isn't measuring the right thing." (OK, maybe in this case not measuring "all the right things." Appreciate your response.
We have a survey question we ask to students annually: “Every adult had 1 or 2 teachers who changed their life. Is your teacher that for you?“ A crude measure, for sure. But that’s the aspiration, and we take the answers seriously.
Fantastic.
> If all of this makes your inner Bayesian flinch, you’re in good company. After twenty‑odd years of watching shiny education fixes wobble and crash—KIPP, AltSchool, Summit Learning, One-laptop-per-child, No child left behind, MOOCs, Khan‑for‑Everything—you should be skeptical.
Wait. Can someone please explain to me how KIPP failed?? AFAIK it's done great for low income minority kids.
KIPP lost its "no excuses" orientation and went a bit "woke" like many other schools but as far as I can tell the academic results are still notably good.
https://www.forbes.com/sites/frederickhess/2020/07/06/kipps-new-mantra-slack-off-be-mean/
I am not familiar with the software used for other subjects, but Duolingo for foreign language instruction for $40k a year is very funny :) I hope you do not expect your children to actually learn any foreign language as you would be disappointed.
I thought that was interesting too. I've used duolingo quite a lot after having learned a few languages the old way previously. It's a really impoverished platform. I started using Pimsleur to compare and it's a much much better program, if you're doing language learning without a live person. But language is a human communication tool -- it seems very odd to me to not learn it through human communication with all the nonverbal cues and emotional involvement that come along with that. To properly learn another language is to absorb another culture and history into your bones, and to some extent to become a somewhat different person in the world.
It feels to me like there are a bunch of deeper layers of learning that happen simultaneously when solving a math problem live with a group of people or discussing history or talking about ways to interpret a piece of writing than can be taught through an algorithm and tested for with multiple choice quizzes. The affordable, scalable two-hour part of this seems much like duolingo to me, which is to say impoverished. And the afternoon enrichment looks much like any other kind of enrichment available to rich private schools.
The good news is that we don’t use Duolingo for foreign language instruction and we agree it doesn’t work. It’s something the students at GT School wanted to try this year.
The bad news is that we don’t have a solution for foreign language instruction at present.
typo that "My 8-year old started 2nd grade content in mid-October 2025" probably meant 2024
> Alpha is much cheaper than a Victorian Governess
Nit: Victorian governesses actually made very little, like £80 ($2000) a year plus room and board. There was very little competition for educated female labor. https://www.whodoyouthinkyouaremagazine.com/feature/governesses
>Workshops in the afternoons are the “fun” part of school. They are the equivalent of the music, theater and art classes that fill in a traditional school schedule (just more focused, measurable and creative).
Am I hopelessly crunchy-granola for thinking that if Alpha is willing to dedicate a whole afternoon program to sports, it would make sense for them to have comparable humanities program? Maybe it’s not “measurable” enough, but engaging with art at a young age seems at least as valuable to me as knowing how to operate an AirBnB.
working on it
I agree. And when it comes to the literature part of the humanities, I don't see how computers can be much help. The thing about the classics is that many of them are hard to access -- hard to find your way into and like. If you can't find your way in -- if Dickens just seems slow and boring, if Shakespeare just seems like a jumble of way way too many unfamiliar words -- then you need someone who loves those writers to show you the way.
There's really no point in forcing yourself to read one of these things if it bores and vexes you from beginning to end, just so you can say you read it. These are pieces of literature that can give immense pleasure. If you can't get over the hump and don't feel any of that, then even if you read it and can answer College Bowl questions about it you haven't read it.
I homeschooled my daughter through 6th grade. Based on my experience, I wanted to give people a data point when it comes to acceleration and speed of learning.
I was extremely laid back regarding book learning until one year before she was going to return to school. By age 5 or she taught herself, with a little help from me, to read and write. (But her spelling was intuitive and phonetic, and her handwriting was very messy.) She picked up a modest amount of add-and-subtract math and simple geometry from games and Montessori type toys. But that was it when it came to the 3 R’s. She did read a lot during her homeschooled years, and was involved in various activities that enriched her mind and life but probably had little effect on testable academic skills.
During the year leading up to her returning to school we worked 2 or 3 hours almost every day on the 3R’s: Spelling, grammar and punctuation, math, handwriting, and expressing ideas in paragraph length bits of writing. I used conventional textbooks and workbooks popular with homeschoolers. Neither of us enjoyed these sessions much, but they worked fine. By the end of the year she was up to speed on all these subjects. I did not attempt to teach her school curriculum science, history, geography etc., since it seemed to me like mostly she would be learning new stuff in those areas, and would not have trouble filling in any gaps in her knowledge if needed.
So she and I sort of did it backward. Instead of getting her way ahead academically, so that she could finish her formal education early and then be free to begin a career or do other things, I let her be free to do other things for 6 years, then taught her the essential skills covered in school in a tiny fraction of the time that would have taken in school. Of course she learned lots of other things during her homeschooled years, and if I wanted to explain more about why I homeschooled the way I did I would describes these. But it’s not relevant to my main point: Teaching the 3 R’s does not take anything like the amount of time schools take with the task. And if you wait til kids are older it far easier for them to learn it — in fact it is extremely easy for them. And my daughter is not one of those intellectually super-gifted kids. In fact when she was small she had some kind of low-grade spoken language disability. (I adopted her from China when she was 2 years old, and language delays and problems are very common in foreign-adopted kids.).
I understand that my entire take on child education is different from the one being described in this review, and am not here to argue about that. Just wanted people to have this data point about how rapidly a kid can master basics, especially when older. You should also know that even the homeschoolers I knew who were trying to approximate a public school curriculum and pace of learning only did a couple hours per day of work on conventional school subjects.
About incentives (bribes). I bribe our dogs so they learn what I want. I use to bribe my kids to go on walks down to the creek. Both with food. I wish someone had given me the idea of bribing my kids to read. I did semi-bribe them by saying I'd buy any book they wanted to read/ look at. "Diary of a Wimpy Kid" got one of my kids hooked on reading that way. I love this review, and fully endorse the idea of doing a non-book review again. Every other year maybe. (I love books too.)
About growing up in an atmosphere that fosters competition.Please do not try to tell me this school does not foster that kind of attitude or that your kids do not partake of it.
<Afternoon programming focused on excelling in “academic competitions” like chess, go, debate, public speaking, robotics, programming and Quiz Bowl.
<This past year four kids placed in the top-8 in a global debate with more than 1000 entries, and two kids are competing at national championships in chess and an academic bee respectively, but not national champions yet.
Come on! Even if your kids don’t get involved in the afternoon “academic competitions” (why put them in quotes? I’d say they’re not “academic competitions” but academic competitions), they are swimming in water with one fat fucking packet of Competitiveness dissolved in it. They’re sucking that solution in through their gills and sending Competitiveness ions up into their brains to have a word with the neuroplasticity supervisor.
I do not doubt at all that a competitive atmosphere has upsides, but I’m here to tell you it has a very substantial downside, too. I’m a psychotherapist and have somehow ended up as one of the go-to CBT therapists in town for super-smart students and young professionals. One person I see was described by the usually down-to-earth neuropsychologist I use for testing as having intellectual powers that are “stratospheric.” There’s somebody who got a perfect score on the LSAT — that’s 99.9 percentile in a test-taking group that’s quite a bit above average intelligence. There’s somebody whose tested IQ when they were high school age was in the 160’s. Unsurprisingly, quite a few of the people are graduate and professional students at the local Ivy.
Some of the supersmarts aren’t Ivy leaguers, though, and were not raised with the expectation that they would be highly successful. And these people come across as having a pretty different take on themselves and on life. There are, for instance, 2 people each of whom had one professional parent and one without a college education. Both were erratic students until they discovered some science and math subjects they loved, then they excelled in those and went on to get degrees in them at places that are not famous, and now have careers in the field. Their point of view about work and achievement is pretty different from the Ivy people’s. They know they are very good at what they do, but they don’t find ways to *signal* that. On the other hand, if you ask them a question that taps into how good they think they are at work skills, they’re cheerfully honest: “Yeah I’m fine at X, Y and Z, but at Q I’m a *monster*! I’ve been playing around with that stuff since I was 16 and I just don’t get stumped by anything anymore. Everybody brings their Q problems to me. And then there’s W. It’s not really in my field, but I was having trouble working with the W guys who did code for some of my projects, so I ended up just learning W myself and doing my own W works out much better. Actually now I’m interested in W itself and have some ideas about a way to revamp it.”
And they don’t seem very prone to jealousy. One works for a company where a management change has brought in a bunch of people who used to work for a couple of the most prestigious tech companies in the world. And he says, “wow. They’re great. They’re some of the smartest and most talented people I’ve ever worked with.” His enthusiasm sounds quite genuine. I can’t tell whether he thinks they’re cleverer than him, but is admiring and entertained, or whether he recognizes that they’re as speedy and clever as he is and are going to be a lot of fun to work with.
Another difference: both have hobbies they love and throw themselves into.
Another difference: neither has much trouble with procrastination.
In contrast, the Ivy Leaguers come across as much more aware, moment-to-moment, of status, whether it’s professional status in the grad program or their job, or who’s making the best quips at dinner. They are vulnerable to intense jealousy, and equally vulnerable to despising themselves for being jealous. They feel deep despair, and might even think about suicide, when forced to realize that somebody is simply more talented than they are.
They have unreasonable expectations of themselves. One became engaged to a woman whose family is mega-wealthy, and felt as though he could not live with himself unless he himself became as rich as the woman’s parents. (He is in a profession where his earnings will make him wealthy, but it will not be wealth on the order of his inlaws’.).
They are very highly motivated to excel, but have trouble with motivation to do the actual work they can excel at.
They are superb networkers, but feel bad about putting more effort into networking than into making new friends and hanging out with their old buddies.
They tend to feel tired and drained after work, and don’t exactly have hobbies — more settings and activities where they socialize and network.
> On Friday the kids are presented with a bunch of locked boxes. One of the boxes has a prize in it. The kids with keys check the boxes to see if their key opens the box with a prize.
Loot boxes. They added loot boxes to kindergarten.
…Okay, fine, that’s not a fair comparison. It’s not a true gambling system because the kids aren’t choosing to spend a resource they could otherwise save up.
But as symbolism goes, it feels like an unflattering metaphor for the system as a whole.
"Once you have freed up half a day for 6-years and a full day for the other seven, you open up a limitless number of possibilities.
Some kids will rush into college classes. Some will choose to use the time to play sports. Some will use the time to master chess or quiz bowl or programming. Some will take time to travel the world with their families."
Sorry to be all grumpy and critical, but I gotta say, that is a *wildly* optimistic and most likely unrepresentative take.
OK, so these kids are limited to those whose parents are both able and willing to spend $40k/year on Alpha, which means. most likely dual-earner couples where both parents have fairly demanding careers. Their kid graduates from Alpha at age 12 instead of 18. Yay! Now what?
"Some kids will rush into college classes." At age 12? Seriously? Are they going to live on campus (in a dorm with a bunch of 18+ year olds? In a frat house?) or are they going to live in their parents' house and commute to and from campus daily, in which case you'd better hope there's good public transit or have a parent who can take you to and from campus daily, because at age 12 you're four years away from being able to get a drivers' license.
Ok, sure, you can do college 100% remotely, but then you miss out on the formative experience of socializing, making friends in person, maybe finding a girlfriend/boyfriend/future spouse.
Some will choose sports! Chess! Programming! Excitement?
Again, how is this going to work *logistically*? Both parents are likely working long hours, probably outside the house. A 12-year-old can be legally left home alone, but cannot drive. Assuming this family lives in the suburbs (= crappy or nonexistent public transit), the kid is stuck home all day. They're not hanging out with other kids their age, because guess what, those kids are at school (Alpha students being a tiny % of the population).
How likely are they to teach themselves chess/programming/other intellectually rewarding pursuits 9 h/day as opposed to, you know, spending the majority of their time on TikTok/video games/p0rn? We're talking *adolescents*, who are not famously known for stellar impulse control, and "intelligent" != "highly motivated autodidact."
Traveling with their family? Sure, that works for maybe 2 weeks/year, i.e., the amount of vacation time the parents have. If you're super lucky, maybe you can send your kid off to travel with a willing grandparent/aunt/uncle for, say, another two weeks, so that covers 1 month out of 12. Also, this works once all your children have graduated from Alpha. If your youngest is still in Alpha or regular school, you can't just leave them behind while you and your firstborn go gallivanting off on a world tour.
TL;DR: This plan has problems, lots and lots of logistical problems.
I haven't seen any comments from Deiseach yet, I eagerly await her appearance with her customary "No shot, Sherlick, of bloody COURSE when you limit yourself to kids whose parents can afford $40k/kid/year for fancy-schmancy schooling, those kids are going to do well! Show me something that works for Tommy from the projects, whose mum is a drug addict living with her fourth boyfriend in the last two years, who also happens to beat her and Tommy."
(No shade on Deiseach; she's excellent at this and I agree with her!)