5 ms·

Class Dismissed: Profile of Joe Liemandt and Alpha School

40 points by surprisetalk 1y ago

skybrian 1y ago
Also see this anonymous entry into an essay-writing contest by one of the parents: https://www.astralcodexten.com/p/your-review-alpha-school https://www.astralcodexten.com/p/your-review-alpha-school
- Animats 1y ago
  Read both. This is fascinating. Notes: - 1 on 1 tutoring yields about 2 standard deviations of improvement, but costs too much. - The Alpha system is very effective but costs about $40,000/student. - It's not that complicated. 2-3 hours a day of intense computer-monitored study, about half using off the shelf tools, and the rest of the day is developed to less structured projects. This is a useful insight. Push kids hard for about 2-3 hours a day and then get to the fun stuff. - Works on ordinary kids. Strong claims in this area. Needs outside validation. - Lots of carrot, including cash payments for achievement. Not much stick. - Staff/student ratio about 1:5. No teachers from bottom quartile of teacher population.
  - renewiltord 1y ago
    Amusingly, $40k is cheap for private school in SF so if they can bring the model here at the same cost...
    - kippinitreal 1y ago
      Their new SF campus charges $75k/yr!
      - renewiltord 1y ago
        Lmao typical. The other schools that charge that are pretty good too.
  - 11101010001100 1y ago
    As noted in the article, for people who won't RTA, the original 2sigma observation of 1 on 1 is due to Bloom more than 40 years ago.
  - alwa 1y ago
    The parent-essayist also mentioned tuition is $10K at the Brownsville TX school, and tuition-free at Alpha’s Arizona charter school.
  - bsder 1y ago
    > - Staff/student ratio about 1:5. No teachers from bottom quartile of teacher population. This is the known formula to accelerate teaching and learning. Even the Gates Foundation (whom I loathe) documented this. Every single one of the successful programs that the Gates Foundation analyzed were declared as too expensive and shut down.
- jeffbee 1y ago
  That article mentions family income as possible confounder only once, while in an article of this extent I would expect it to appear dozens of times and be the central point. You simply cannot draw conclusions from comparing general populations and kids in schools that cost $500k for a 13-grade education, or that cost any amount. Students from low income households have so many headwinds. They don't come to school because their transportation is unreliable, because their teeth hurt, because they can't afford sanitary pads, they can't sleep because their neighborhoods are too loud, and so on. When this school opened in Brownsville and preferentially enrolled only the children of employed aeronautical engineers, that precluded any possible comparison of outcomes.
  - skybrian 1y ago
    Yes, they are tiny and it has to be considered experimental and unproven as far as scalability goes. They are a long way from trying to educate millions of kids. But it will be interesting to see how far it scales up and into what demographics before it stops working and they reach their limits. So far, it seems they've had success with in-person schools and it didn't work (yet) remotely. Motivation seems like the key factor.
    - underlipton 1y ago
      Scalability is the issue, though, isn't it? Any number of esoteric pedagogical approaches have been tested, to great success, with the children of people wealthy enough to fulfill all of their scions' material needs and subsequently move onto purchasing immaterial things like "attention". You have to figure out something you can apply kids in spite of who their parents are. If not, you have to admit that there's no such thing as a meritocracy divorced from what the parents are willing or able to provide (which opens up questions of how much children deserve to have that deficit filled, it if happens to exist in any individual case).
      - skybrian 1y ago
        I don't know anything about private schools beyond reading a few articles, but I think that might be overstated. Even with the same kids and parents, it seems likely that some schools are better than others?
azrgll 1y ago
[dead]
didibus 1y ago
I'd be curious to have adults use their learning program as well, and assess their aptitude at learning with it.
neilv 1y ago
Warning: The article is ~18K words. Longer than a long-form journalism article; more like novella length. I started skimming faster when it diverted into an extended YA drama, and eventually gave up.
- sam1r 1y ago
  Yeah, I almost thought I accidentally clicked to a completely different article. A couple of whole pages scrolled later, I was like oh.. ok. This is how this connects. Feels like a film script.
underlipton 1y ago
Quick anecdote before I dive in: >learning things like how to run an Airbnb or food truck, manage a brokerage account or Broadway production, or build a business or drone. I remember touring my high school alma mater, a local magnet with a decent record of getting kids into top schools (that it was fed by the bedroom communities of NASA and USDA sites clearly notwithstanding). As we passed the shop, we spent a good 15 minutes learning about the small plane students were building. A good 10-foot wingspan. I matriculated and that was the last I heard of it. I've noticed that, in marketing for educational institutions, it's common for the uncommon and exceptional achievements of That One Student to get pushed as indicative of what most students can expect to (and will never actually) do. I'd think most would do well to be wary of an article that starts out with such a claim. Regression to the mean probably factors in, even for outlier schools.
- AtlasBarfed 1y ago
  And he's an AI guy pushing AI things in an AI hype cycle. That said, some combination of AI + Wikipedia could probably do amazing things for teaching a student. Here's the thing: if the student trusts that he can ask any stupid question to an AI, a well trained one can probably walk students through concepts in any of thousands of different questions asked by previous students and lead them through to the end goal. I can see the potential. But these guys? No. Education should be humanist, not capitalist. You can already see it: look at the little entrepreneur parrots we created. This isn't education to benefit the child: it's marketing to the parents to get them to pay up because their precious child will become the next Bill Gates.
  - kikimora 1y ago
    He is not AI guy. SalesBuilder is a constraint solver, not AI, not even an expert system.
    - skybrian 1y ago
      From the article though, Joe Liemandt and an AI enthusiast now and funding a lot of work on educational software assuming AI is the answer. The software alone seems not to be the answer, though.
llm_nerd 1y ago
Liemandt was on the Peter Attia "The Drive" podcast a couple of days ago discussing this. The episode is 1 hr 37 minutes before you start applying multipliers.
- sam537 1y ago
  I came to say this. Free episode to listen to. If you are interested in this it is Nx more cost-effective to listen to Liemandt than to read the adjective laden article.

40 points by surprisetalk 1y ago

Also see this anonymous entry into an essay-writing contest by one of the parents: https://www.astralcodexten.com/p/your-review-alpha-school https://www.astralcodexten.com/p/your-review-alpha-school

Animats 1y ago
Read both. This is fascinating. Notes: - 1 on 1 tutoring yields about 2 standard deviations of improvement, but costs too much. - The Alpha system is very effective but costs about $40,000/student. - It's not that complicated. 2-3 hours a day of intense computer-monitored study, about half using off the shelf tools, and the rest of the day is developed to less structured projects. This is a useful insight. Push kids hard for about 2-3 hours a day and then get to the fun stuff. - Works on ordinary kids. Strong claims in this area. Needs outside validation. - Lots of carrot, including cash payments for achievement. Not much stick. - Staff/student ratio about 1:5. No teachers from bottom quartile of teacher population.
- renewiltord 1y ago
  Amusingly, $40k is cheap for private school in SF so if they can bring the model here at the same cost...
  - kippinitreal 1y ago
    Their new SF campus charges $75k/yr!
    - renewiltord 1y ago
      Lmao typical. The other schools that charge that are pretty good too.
- 11101010001100 1y ago
  As noted in the article, for people who won't RTA, the original 2sigma observation of 1 on 1 is due to Bloom more than 40 years ago.
- alwa 1y ago
  The parent-essayist also mentioned tuition is $10K at the Brownsville TX school, and tuition-free at Alpha’s Arizona charter school.
- bsder 1y ago
  > - Staff/student ratio about 1:5. No teachers from bottom quartile of teacher population. This is the known formula to accelerate teaching and learning. Even the Gates Foundation (whom I loathe) documented this. Every single one of the successful programs that the Gates Foundation analyzed were declared as too expensive and shut down.
jeffbee 1y ago
That article mentions family income as possible confounder only once, while in an article of this extent I would expect it to appear dozens of times and be the central point. You simply cannot draw conclusions from comparing general populations and kids in schools that cost $500k for a 13-grade education, or that cost any amount. Students from low income households have so many headwinds. They don't come to school because their transportation is unreliable, because their teeth hurt, because they can't afford sanitary pads, they can't sleep because their neighborhoods are too loud, and so on. When this school opened in Brownsville and preferentially enrolled only the children of employed aeronautical engineers, that precluded any possible comparison of outcomes.
- skybrian 1y ago
  Yes, they are tiny and it has to be considered experimental and unproven as far as scalability goes. They are a long way from trying to educate millions of kids. But it will be interesting to see how far it scales up and into what demographics before it stops working and they reach their limits. So far, it seems they've had success with in-person schools and it didn't work (yet) remotely. Motivation seems like the key factor.
  - underlipton 1y ago
    Scalability is the issue, though, isn't it? Any number of esoteric pedagogical approaches have been tested, to great success, with the children of people wealthy enough to fulfill all of their scions' material needs and subsequently move onto purchasing immaterial things like "attention". You have to figure out something you can apply kids in spite of who their parents are. If not, you have to admit that there's no such thing as a meritocracy divorced from what the parents are willing or able to provide (which opens up questions of how much children deserve to have that deficit filled, it if happens to exist in any individual case).
    - skybrian 1y ago
      I don't know anything about private schools beyond reading a few articles, but I think that might be overstated. Even with the same kids and parents, it seems likely that some schools are better than others?

azrgll 1y ago

[dead]

didibus 1y ago

I'd be curious to have adults use their learning program as well, and assess their aptitude at learning with it.

neilv 1y ago

Warning: The article is ~18K words. Longer than a long-form journalism article; more like novella length. I started skimming faster when it diverted into an extended YA drama, and eventually gave up.

sam1r 1y ago
Yeah, I almost thought I accidentally clicked to a completely different article. A couple of whole pages scrolled later, I was like oh.. ok. This is how this connects. Feels like a film script.

underlipton 1y ago

Quick anecdote before I dive in: >learning things like how to run an Airbnb or food truck, manage a brokerage account or Broadway production, or build a business or drone. I remember touring my high school alma mater, a local magnet with a decent record of getting kids into top schools (that it was fed by the bedroom communities of NASA and USDA sites clearly notwithstanding). As we passed the shop, we spent a good 15 minutes learning about the small plane students were building. A good 10-foot wingspan. I matriculated and that was the last I heard of it. I've noticed that, in marketing for educational institutions, it's common for the uncommon and exceptional achievements of That One Student to get pushed as indicative of what most students can expect to (and will never actually) do. I'd think most would do well to be wary of an article that starts out with such a claim. Regression to the mean probably factors in, even for outlier schools.

AtlasBarfed 1y ago
And he's an AI guy pushing AI things in an AI hype cycle. That said, some combination of AI + Wikipedia could probably do amazing things for teaching a student. Here's the thing: if the student trusts that he can ask any stupid question to an AI, a well trained one can probably walk students through concepts in any of thousands of different questions asked by previous students and lead them through to the end goal. I can see the potential. But these guys? No. Education should be humanist, not capitalist. You can already see it: look at the little entrepreneur parrots we created. This isn't education to benefit the child: it's marketing to the parents to get them to pay up because their precious child will become the next Bill Gates.
- kikimora 1y ago
  He is not AI guy. SalesBuilder is a constraint solver, not AI, not even an expert system.
  - skybrian 1y ago
    From the article though, Joe Liemandt and an AI enthusiast now and funding a lot of work on educational software assuming AI is the answer. The software alone seems not to be the answer, though.

llm_nerd 1y ago

Liemandt was on the Peter Attia "The Drive" podcast a couple of days ago discussing this. The episode is 1 hr 37 minutes before you start applying multipliers.

sam537 1y ago
I came to say this. Free episode to listen to. If you are interested in this it is Nx more cost-effective to listen to Liemandt than to read the adjective laden article.