3 ms·

> Thanks for the very thoughtful response! I admit I also get salty, particularly when it comes to education. It feels so obvious things should be a specific wa

> Thanks for the very thoughtful response! I admit I also get salty, particularly when it comes to education. It feels so obvious things should be a specific way, but of course that may just be my STEM person arrogance :P. Thank you (also for indulging)! As an also-arrogant STEM person myself we can muddle through together haha. > So, what I'm mostly confused about is why expulsion wouldn't work. I think a number of dynamics are at play here: - Schools don't usually reach for suspension/expulsion that quickly because they're weighing the impact of the problem kid's behavior on others vs. the impact of a suspension/expulsion on the kid, so their disruptive lingers. - Some schools have zero tolerance policies that suspend/expel very quickly, but it turns out that creates a super weird climate (students defending themselves are also suspended/expelled, school staff feel pretty bad suspending/expelling all the time, you can't build relationships with problem kids which is deeply dehumanizing on both sides, etc.) - Problem kids have a weird habit of just coming right back. A lot of us are envisioning a relatively rich school district with multiple nets to cordon off problem kids, bost districts have the one school, maybe if they're lucky there's an "alternative school" in the parking lot, which is a trailer that should only ever have 5 people in it, but it has 15. Maybe some people are advocating for some kind of super harsh zero-tolerance-expelled-forever pipeline, but let me introduce those advocates to the School-to-Prison Pipeline [0]. - Problem kids are still in your neighborhood, your kid is pretty likely to still see them outside of school, and that leads to more weird social dynamics. But moreover, let's say that zero-tolerance-expel-immediately leads to better outcomes for kids and we have some way of totally segregating problem kids both in school and broader society. Those kids are still a problem for society that we'll have to deal with at some point. Today they're throwing stuff in class, tomorrow they're breaking the windows of your car or running drugs in your neighborhood. At that point in the School-to-Prison Pipeline, rehabilitating the person is extremely expensive. > seems to be a clear-cut case of Simpson's paradox Nah, definitely not. A commonly cited paper [1] has a pretty good table breaking down the effects of various classroom properties on outcomes. Reading it, you'll immediately get a great look at why private/charter school outcomes are so much better: they work pretty hard to cherry-pick kids that lead to better outcomes, thus exacerbating the School-to-Prison Pipeline issue by putting more pressure on public schools. Anyway, there's so much on this topic you're gonna have to switch your argument to explaining a conspiracy in educational research: Suspending Progress: Collateral Consequences of Exclusionary Punishment in Public Schools: https://sci-hub.st/https://doi.org/10.1177/0003122414556308 https://sci-hub.st/https://doi.org/10.1177/0003122414556308 Effects of School-Wide Positive Behavioral Interventions and Supports on Child Behavior Problems: https://pmc.ncbi.nlm.nih.gov/articles/PMC3483890/ https://pmc.ncbi.nlm.nih.gov/articles/PMC3483890/ Teacher Support for Zero Tolerance Is Associated With Higher Suspension Rates and Lower Feelings of Safety: https://sci-hub.st/https://doi.org/10.1080/2372966X.2020.1832865 https://sci-hub.st/https://doi.org/10.1080/2372966X.2020.183... Schoolwide positive behavioural interventions and supports and human rights: transforming our educational systems into levers for social justice: https://pmc.ncbi.nlm.nih.gov/articles/PMC9897741/ https://pmc.ncbi.nlm.nih.gov/articles/PMC9897741/ School-wide positive behavioral interventions and supports and students with extensive support needs: a scoping review: https://pmc.ncbi.nlm.nih.gov/articles/PMC9897773/ https://pmc.ncbi.nlm.nih.gov/articles/PMC9897773/ The Impact of Schoolwide Positive Behavioral Interventions and Supports on Bullying and Peer Rejection: https://jamanetwork.com/journals/jamapediatrics/fullarticle/1107694 https://jamanetwork.com/journals/jamapediatrics/fullarticle/... > Stuff about blame Blame essentially never works, and it's because people are the products of systems. You talk about Goodhart's Law; another dynamic is where we do things that feel good or confirm our understanding of the world despite poor outcomes. Harsh disciplinary policies are the poster child for this. I'm gonna assume here you're pretty naive to the criminal justice space (this is because anyone who knows anything about criminal justice understands blame essentially never works), so I strongly encourage you to interrogate your priors here and read up on deterrence, punishment, and so on. > National placement test for each grade This would really only measure socioeconomic status, like most (all?) standardized tests. You also get stuck in tracks, so if say your mom dies in 3rd grade, you do poorly on the test and get bumped down, you're probably bumped down forever. That's a bad outcome. > Disruptive students get kicked to penitentiaries. Not only are there completely valid reasons for students becoming disruptive (parental issues, injuries, mental health issues, etc), the expense of this is out of this world. Even in the cheapest state (Arkansas) spending-per-inmate is $23k/yr--the median is something like $60k. Your options here are dramatically increase taxes or create a truly horrific human rights disaster. > The same people that write olympiad problems and run the summer camps are hired to create a new curriculum. This doesn't work because different people learn in different ways. You need dedicated, educated, well-compensated, supported professionals applying state-of-the-art techniques and research to get the outcomes we want. Also when you talk about replacing Common Core with some new standard, you're still not escaping Goodhart's Law. > Everyone is fired, and as many people as money there is are hired at $300–500k/yr (in total compensation) to teach [30 student classes]. 30 student classes obviates any benefit you'd get from anything else. There are no systems with those class sizes that are achieving the outcomes we want. You also can't literally fire all teachers. The NEA or teacher tenure won't let you. Training and interviewing 2.333 million teachers (plus administrators) is a gargantuan undertaking. Who moves to the middle of nowhere in Arizona, or Mississippi? How will you find so many qualified people? That $15k/yr number you keep citing isn't all salary; we spend around $236b on ~4m teacher salaries, which yields ~600k teachers (at $400k/yr salary), so you still need to find $680b (which is more than the budget of Medicaid) for the remaining 1.7 million teachers. You also have to somehow survive the political fallout of firing hundreds of thousands of people, many of whom have families and various health issues. [0]: https://en.wikipedia.org/wiki/School-to-prison_pipeline https://en.wikipedia.org/wiki/School-to-prison_pipeline [1]: https://pmc.ncbi.nlm.nih.gov/articles/PMC9159706/#T2 https://pmc.ncbi.nlm.nih.gov/articles/PMC9159706/#T2

programjames 2y ago
I think the school->prison pipeline is a real issue, but I think a poor quality of education is a much bigger deal because smart, educated people generate exponentially more wealth since the industrial revolution. If you want what is best for everyone, you would focus more resources on top-performing students rather than less! Sure, top-performing students would turn out better than mid-performing students—even with fewer resources—but that's a tautology and an emotional appeal. I think the tricky part is to make sure top students give back to society once they graduate, but that seems more of a cultural issue to solve. Boring students to death probably doesn't help, though. Now, you brought up that national testing + placement would mostly reflect socio-economic status. I think this is concerning because it lead to in-groups reinforcing themselves, which naturally decreases motivation for future rich people to help the rest of society. However, we already have examples of placement tests, and this isn't what happens! NYC has several "specialized" schools, including one of the best high schools in the nation, Stuyvesant. Admissions to Stuyvesant are entirely based on your rank on the SHSAT, yet 48% of their students are "economically disadvantaged" according to USNews. I'm not entirely sure what that means, but I would expect lower or lower-middle class. This data also matches up with my intuitions: although intelligence is heritable (through genes or upbringing), there are exponentially more "economically disadvantaged" people than rich people, so even though rich kids are overrepresented, they are still outnumbered by poor(er) kids. Also, keep in mind that rich people will always be able to pay for private schools or tutors if they find public education lacking. So, you are really only depriving poor students of any possibility of a good education by lumping everyone together, which is worse for reinforcing classism. As you mentioned, charter/private school outcomes are so much better because they work pretty hard to cherry-pick kids that lead to better oucomes. Why not give everyone that opportunity? That's a little facetious, because not everyone has that opportunity. Some people are just not genetically predisposed towards exams, or they're being abused at home, or they have to work after school to buy food for their younger siblings. But, it doesn't really matter why someone cannot do/be better if we're unable to fix the why. Until it can be fixed, the problem is just a part of them and they'll be punished for it. This isn't very sympathetic, but it's the game-theoretical optimal approach for getting to the Pareto frontier. You mention that blame/punishment essentially never works, which is probably because humans are not perfectly rational agents. Sure. I've definitely seen this when I play Risk online. You have to use different strategies when people are irrational/prone to mistakes, e.g. with novices it's usually good to make a big stack and wait for everyone else to noob-slam, while with masters it's better to work with the othe rplayers to slowly choke out the rest. Optimal strategies may be less tolerant to mistakes, and a common mistake humans make is, "this person hurt me, so I will hurt them even more," without considering why they were hurt. A common theme I saw in school->prison pipeline studies is that youth get disaffected with society/the justice system, so they end up committing more crimes. If people really are being irrational, in such a way that punishment will not work, you really only have three options: 1. Force them into rationality. 2. Rehabilitate them through positive reinforcement. 3. Eliminate them from society, e.g. sending them to Louisiana/Australia, prisons/executions, or closed communities. I'd argue that you should take whichever option is best for society, i.e. costs it the least. Why? a) Societies cannot be comprised of mostly (weighting by utility) negative-externality people for very long. b) Everyone else is better off by eliminating such people, thus they are motivated to do so in whichever way is cheapest. If it were cheaper to just execute all criminals, or commit horrific human rights disasters to make prisons cheap to run, that's what society should do. Historically, that's what societies have done. Nowadays, it probably isn't cheaper; even if the average inmate spends just as much time in prison as out of it, they are probably close to net-positive to society. The cheapest solution probably is rehabilitation for most people except the unfixable, and even there, life in prison is probably cheaper than execution. So, I think I agree with you about rehabilitation, but probably not for the reasons you cite. I think "people being products of systems" is a rather naive take; if people were products of stable systems (in the physics sense), punishment actually would work. Conservatives have a bias towards everything being a stable system [which is true; you are exponentially more likely to end up in more stable (determined by transition probabilities) trembling-hand/thermodynamic equilibria], which is probably why they're all pro-punishment and such. Note that rehabilitation can still be cheaper, but at least punishment would work. It's only when you have unstable systems that punishment might not work at all. It's a little worrying to think that America's system might be unstable right now, but the race riots and past two elections kind of show it is. More accurately, it's too easy to transition out of its current maximum for punishment to really dissuade future malcontents. ----- Alright, let's return to education. I think we're in agreement that: I) The school->prison pipeline is real. II) Imprisonment is expensive, probably moreso than rehabilitation. As I put at the top, we probably disagree that: III) This cost is more than that of a poor education system. I think rehabilitation through the education system is far more expensive. Here's just a back-of-the-napkin calculation. Suppose that all inmates are directly a result of the school->prison pipeline. It costs ~$70bn/yr to incarcerate them, but let's also assume we're missing out on 2 million people * $65k/yr = $130bn from jobs they could be working. This amounts to about $200bn/yr in costs to society. Now, the number of billionaires increases by about 36 each year, and the average billionaire has $7bn. If a better education were to double the number of billionaires produced each year, this would entirely offset the cost. Of course, billionaires are usually better at capturing value than producing it, but at least some educated STEM guy below them is producing the value. I think this is entirely doable by expelling more students. In fact, I think the justice system will only start costing more than the wealth generated from better education (through the top students) once the pitchforks and torches come out. Similarly, we should be able to do a cost-benefit analysis on teacher salaries. In reality, the most efficient use of money is to pour it into the top schools (and have entrance exams), but the metrics don't look as good. Most legislators care about the quintiles, not the top 0.1%, plus the wokists hate inequity. And, even bottom schools need much better teachers than are currently around. That's why I want to raise all salaries to $300-500k/yr. You mentioned that only $236bn is spent on ~4m teachers, but there actually is another $600bn going elsewhere. If you want an average class size of 20 students, you only need ~2.5m teachers, so it should be possible if you strip everything else to the bare-bones. My elementary school class was taught in a portable, and I think that's better than having a shitty teacher.

programjames 2y ago

I think the school->prison pipeline is a real issue, but I think a poor quality of education is a much bigger deal because smart, educated people generate exponentially more wealth since the industrial revolution. If you want what is best for everyone, you would focus more resources on top-performing students rather than less! Sure, top-performing students would turn out better than mid-performing students—even with fewer resources—but that's a tautology and an emotional appeal. I think the tricky part is to make sure top students give back to society once they graduate, but that seems more of a cultural issue to solve. Boring students to death probably doesn't help, though. Now, you brought up that national testing + placement would mostly reflect socio-economic status. I think this is concerning because it lead to in-groups reinforcing themselves, which naturally decreases motivation for future rich people to help the rest of society. However, we already have examples of placement tests, and this isn't what happens! NYC has several "specialized" schools, including one of the best high schools in the nation, Stuyvesant. Admissions to Stuyvesant are entirely based on your rank on the SHSAT, yet 48% of their students are "economically disadvantaged" according to USNews. I'm not entirely sure what that means, but I would expect lower or lower-middle class. This data also matches up with my intuitions: although intelligence is heritable (through genes or upbringing), there are exponentially more "economically disadvantaged" people than rich people, so even though rich kids are overrepresented, they are still outnumbered by poor(er) kids. Also, keep in mind that rich people will always be able to pay for private schools or tutors if they find public education lacking. So, you are really only depriving poor students of any possibility of a good education by lumping everyone together, which is worse for reinforcing classism. As you mentioned, charter/private school outcomes are so much better because they work pretty hard to cherry-pick kids that lead to better oucomes. Why not give everyone that opportunity? That's a little facetious, because not everyone has that opportunity. Some people are just not genetically predisposed towards exams, or they're being abused at home, or they have to work after school to buy food for their younger siblings. But, it doesn't really matter why someone cannot do/be better if we're unable to fix the why. Until it can be fixed, the problem is just a part of them and they'll be punished for it. This isn't very sympathetic, but it's the game-theoretical optimal approach for getting to the Pareto frontier. You mention that blame/punishment essentially never works, which is probably because humans are not perfectly rational agents. Sure. I've definitely seen this when I play Risk online. You have to use different strategies when people are irrational/prone to mistakes, e.g. with novices it's usually good to make a big stack and wait for everyone else to noob-slam, while with masters it's better to work with the othe rplayers to slowly choke out the rest. Optimal strategies may be less tolerant to mistakes, and a common mistake humans make is, "this person hurt me, so I will hurt them even more," without considering why they were hurt. A common theme I saw in school->prison pipeline studies is that youth get disaffected with society/the justice system, so they end up committing more crimes. If people really are being irrational, in such a way that punishment will not work, you really only have three options: 1. Force them into rationality. 2. Rehabilitate them through positive reinforcement. 3. Eliminate them from society, e.g. sending them to Louisiana/Australia, prisons/executions, or closed communities. I'd argue that you should take whichever option is best for society, i.e. costs it the least. Why? a) Societies cannot be comprised of mostly (weighting by utility) negative-externality people for very long. b) Everyone else is better off by eliminating such people, thus they are motivated to do so in whichever way is cheapest. If it were cheaper to just execute all criminals, or commit horrific human rights disasters to make prisons cheap to run, that's what society should do. Historically, that's what societies have done. Nowadays, it probably isn't cheaper; even if the average inmate spends just as much time in prison as out of it, they are probably close to net-positive to society. The cheapest solution probably is rehabilitation for most people except the unfixable, and even there, life in prison is probably cheaper than execution. So, I think I agree with you about rehabilitation, but probably not for the reasons you cite. I think "people being products of systems" is a rather naive take; if people were products of stable systems (in the physics sense), punishment actually would work. Conservatives have a bias towards everything being a stable system [which is true; you are exponentially more likely to end up in more stable (determined by transition probabilities) trembling-hand/thermodynamic equilibria], which is probably why they're all pro-punishment and such. Note that rehabilitation can still be cheaper, but at least punishment would work. It's only when you have unstable systems that punishment might not work at all. It's a little worrying to think that America's system might be unstable right now, but the race riots and past two elections kind of show it is. More accurately, it's too easy to transition out of its current maximum for punishment to really dissuade future malcontents. ----- Alright, let's return to education. I think we're in agreement that: I) The school->prison pipeline is real. II) Imprisonment is expensive, probably moreso than rehabilitation. As I put at the top, we probably disagree that: III) This cost is more than that of a poor education system. I think rehabilitation through the education system is far more expensive. Here's just a back-of-the-napkin calculation. Suppose that all inmates are directly a result of the school->prison pipeline. It costs ~$70bn/yr to incarcerate them, but let's also assume we're missing out on 2 million people * $65k/yr = $130bn from jobs they could be working. This amounts to about $200bn/yr in costs to society. Now, the number of billionaires increases by about 36 each year, and the average billionaire has $7bn. If a better education were to double the number of billionaires produced each year, this would entirely offset the cost. Of course, billionaires are usually better at capturing value than producing it, but at least some educated STEM guy below them is producing the value. I think this is entirely doable by expelling more students. In fact, I think the justice system will only start costing more than the wealth generated from better education (through the top students) once the pitchforks and torches come out. Similarly, we should be able to do a cost-benefit analysis on teacher salaries. In reality, the most efficient use of money is to pour it into the top schools (and have entrance exams), but the metrics don't look as good. Most legislators care about the quintiles, not the top 0.1%, plus the wokists hate inequity. And, even bottom schools need much better teachers than are currently around. That's why I want to raise all salaries to $300-500k/yr. You mentioned that only $236bn is spent on ~4m teachers, but there actually is another $600bn going elsewhere. If you want an average class size of 20 students, you only need ~2.5m teachers, so it should be possible if you strip everything else to the bare-bones. My elementary school class was taught in a portable, and I think that's better than having a shitty teacher.