This is the sentence I can't get past: "listen to a teacher explain the same material at the same pace in the same way". I don't doubt that many people experienced school that way, or even that there are plenty of classrooms where that is the primary mode of instruction. But the belief that that kind of classroom environment is the best the field of education has to offer is really difficult for me to fathom because if you picked up one book or talked to a single expert it would immediately disabuse you of that belief. I think the relevant question is not "How can we use technology to personalize learning for students?" but "How can we spread 'personalized' learning experiences to more classrooms?" (like the type Dan describes here) which is more of a societal question than a technological one. But tech billionaires only have a hammer and that makes everything a nail.
Another, wonderful example of authentic teaching and learning mathematics. A real problem solved in a community of math curious kids with a teacher who underscores understanding her students' thinking... The public nature of the work and collaboration is so important. Thank you Dan for sharing this.
Amen!! I just wrote something (https://jendycksprout.substack.com/p/the-best-education-is-not-in-our ) along the lines of what you are arguing here and a reader directed me to your piece which has me clapping at my desk. Especially loved how you summarized the problem in this paragraph:
"The idea that computers should personalize instruction, flattening the human differences that learners would much rather see celebrated and developed, maintains an absolute stranglehold on the imagination of the billionaires who fund education technology. Small children will touch a hot stove only once yet billionaires will fund personalized learning initiatives again and again and again and we might wonder why. What accounts for the appeal of this idea? "
I have been reading your blogs for the last year or so and feel so inspired by them as a new math teacher. I believe in group learning and hear my students complain about feeling isolated in personalized classes all the time. This year I am working on designing a gifted education math program in a school that already exists on a mastery learning module. When I bring up the cons of personalized learning to the school leaders, they argue that having such a predesigned curriculum is helpful because not all teachers are trained or experienced enough to have a class that can be successful otherwise. We do live in a place where there is hardly any teacher training and if there is, its not sufficient at all, and yet I do not believe that our only solution is personalized mastery learning courses but I am not sure what to offer instead. Do you have any thoughts?
Thanks for another great piece, Dan - you've been on fire lately. (My lifelong educator mom agrees - she excitedly emailed me this post this morning).
As someone who's also spent time at Summit, I have to point out that you're actually in violent agreement with much of their model from ~2015. They positioned creative group work, led by an excellent teacher like Liz, as the place where the most important learning happened. I actually remember observing a lesson pulled out of your playbook - something to do with creating graphs in Desmos to match an observed phenomenon.
To me, Summit's model accepted as necessary that kids needed "fundamentals" - to acquire a bunch of knowledge and do a bunch of repetitive practice, and the best & most efficient way to get that done was with headphones, laptops, and student independence. I imagine you'd argue that doing that kind work outside of an engaging & meaningful context is ineffective. And the results from Summit Learning may prove you right – maybe any attempt to separate "rote learning" and "meaningful learning" is doomed to failure.
Summit may be catching some strays here. We have hired excellent people from Summit. They were early test sites for our curriculum. (The observed phenomenon graph you mentioned.) If Zuckerberg were investing in the 24 hours of the school week Summit students spend in whole group instruction, or even the full 40, this would be a different conversation. But that 16 hours of locked-into-laptop time seems particularly appealing to billionaire funders in ways that a) are interesting to me, and b) consistently underperform expectations.
I guess the proto-question here which I haven't answered is - when / how much should students work on laptops by themselves, and doing what?
As someone who also spent time with Summit, I'd also suggest that their philosophy is the same as yours. Your example of Liz's sample class is a model of what Summit would want to spend time with. It's separating conceptual understanding from procedural fluency. The online mini-quizzes focusing on process, left me more lesson time to devolop lessons that anticipated, monitored, selected, sequenced and connected student thinking. It was Summit PD that introduced me to that routine. As you know, your 3 Act Tasks were all over the Summit platform.
The model I learned at Summit was not students parked behind a laptop. Even for the 10 pt quiz time, those vilified 16 hours (have you been to classrooms post pandemic? In our school, students don't even vbring paper to class anymore, my class is often the onlyone they need to bring a pencil), we were encouraged to use the data for mini-lessons offline.
I'd suggeat that blaming "out of touch" billionaires is the easy way out of explaining Summit's failure.
What I saw as the fundamental challenge for our community is the same thing that challenges any kind of math reform efforts. It's not like they learned it. Parents, teachers, and administrators alike don't understand this complicated relationship between conceptual knowledge vs. procedural fluency. Students who had been praised for being able to just "do" the procedures (ace the quizzes) didn't see the value in conceptual lessons. It was easy to blame billionaires, rather than opening themselves to a deeper learning experience.
The fact that these happened to be liberal billionaires, and that the curricula happened to include math lessons on identifying racial profiling, questioning the role of folks like Rockefeller and Columbus, made for easy pickings fror our conservative, rural town.
Also, the ambitious administrator and former math teacher who spearheaded the adoption of Summit over the summer resigned the firat day of school, leaving an unsupportive building principal, and a handful of teachers like me who were put up to absorb all of the community's frustration.
As a school, I also don't think we executed the curriculum as intended, and for most classes 100% of the time was behind the laptop. Students would tell me that I was the only one still doing paper and pencil. I have great colleagues, but we were attacked from the beginning, and mostly left to fend for ourselves.
My point here is that I think your analysis on Summits failure throws away the baby with the bathwater. I don't like billionaires any more than you, but the Summit product was good enough for your name, and Desmos to be all over it.
Are you ready to give up on the idea that conceptual understanding and procedural fluency are both important? Sure, lets build teachers focused on assets, not deficits - that seems totally compatible with Summit.
I wouldn't go back to Summit, but the philosophy (if not execution) seems directly in line with what I've been reading from you for nearly 15 years.
Lastly, in answer to your new post about what would you do with $100M, have you considered using it to leverage $100B in annual federal funding to lighten the load on teachers so that weathy techies aren't tempted to plug kids into their products?
" "listen to a teacher explain the same material at the same pace in the same way".
I think this is the disconnect. If you're Zuck or Gates, forced to spend time in a classroom with an inferior teacher with nothing to offer such fine minds as theirs, all they want to do is acquire the information themselves. Education isn't, for them, a social act. They either want all the teacher's attention spent on the "smart" kids or the teacher left for the unworthy, letting the superior kids learn at their own pace, far past the one that mere mortals can manage.
They aren't saying that *now*, of course. I think they are genuinely trying to help. But what they remember from their own classrooms is their own boredom and the wide range of abilities. They never saw school as a social event or interaction.
In that context, it's significant that they are both college dropouts. They didn't want to learn what they didn't want to learn, and there was no value in the social action.
I think this is a valid critique and an important conversation, but you’re not really engaging with the problem as envisioned by these philanthropic non-educators.
In their view, it is a fundamental problem that all of the students are working on pool border problems at the same time because “that’s what on the schedule.” It’s inefficient and maybe even unjust that kids that already understand the algebra involved have to stay and twiddle their thumbs, and kids that aren’t ready yet have to fake it, because that’s what the logistical constraints of the classroom demand.
Education outsiders observe (I think correctly) that a universal timeline for all learners is really wasteful. We don’t do this because it’s good for learning, we do this because it’s cheap. I think most people would agree that if kids were in much smaller classrooms or received individual tutoring that was targeted to their level, they’d learn a lot more and/or experience much less frustration and grief at being left behind.
The thesis then is that maybe digital technology would make it economical to allow each student to have their own timeline, or at very least, to have some variation based on their ability and prior knowledge. This usually goes poorly because solutions under this thesis tend to ignore all of the social elements of learning that you’re alluding to, because they don’t scale well.
Thanks for the thoughtful response here, Zane. Point of agreement: our current model of large groups of age-graded children accomplishes two goals—one a goal of convenience, as you mention, and the other a goal of socialization. I have not seen a model, especially a computer-based one, that disentangles those goals without great cost.
A point of disagreement, in case you have any follow-up thoughts -
> It’s inefficient and maybe even unjust that kids that already understand the algebra involved have to stay and twiddle their thumbs, and kids that aren’t ready yet have to fake it, because that’s what the logistical constraints of the classroom demand.
My first thought is that every kid is ready to say what they noticed about the pool that Liz flashed on the screen. Every kid is ready to count up the tiles at minimum and many are ready to think about the tile count in more abstract ways. So I want to reject the idea that the floor in math class must necessarily be above the reach of groups of kids even in situations with multi grade levels of learner variability. And I also want to reject the idea that the ceiling is as low as you describe.
I say this as someone who picked up a couple of textbooks of advanced math and skipped from one math class into the next over a two-week holiday break: I knew that math up to a certain level but I didn't understand its depth or connectivity in the least and I would have benefited from someone pushing me to develop my understanding much more.
For example, in this problem, we could ask a student to generalize their solution from an n x n square to a n x m rectangle. We could ask them how an extra layer of tiles would change their solution. We could give them a number of tiles and ask them to tell us the dimensions of the pool, instead of the inverse.
I just want to check what feels like a latent assumption in lots of these conversations, that in a class of 30 kids, only one or two of those kids can get their needs met at max. That isn't the reality in lots of classrooms, even ones with lots of learner variability.
Heartily agree that socialization is a worthy goal of classroom learning, at least for me.
I love the idea of varying the difficulty/complexity of the same “problem” via elaboration, complications, etc. This resonates deeply with some lessons from a past life working on world language learning, where I was taught, “change the task, not the text.” I.e. use authentic content that may not be fully comprehensible to all learners, but adapt by not asking them to fully comprehend yet (instead estimate the gist, or enumerate context clues, or recognize key words, etc). Two very differently leveled learners could both benefit from the same content.
I think this is a lot more promising than “every kid on their own, in parallel,” but the skeptic in me wonders how much this relies on the heroics of an individual teacher to surmount structural hurdles. It takes a lot of extra time to elaborate level variations within a topic/learning outcome/problem. And then at the end of the day/unit will we still assess students uniformly? I’m here for learning for its own sake, even if it never shows up on a test, but when push comes to shove anything that’s not on the test tends to get cut first, right?
Then I think about how variation might compound over months and years. In terms of the granularity of the variation, you’re looking at an individual problem/lesson. Which is awesome! But at some point doesn’t that naturally mean that some learners ought to be on totally separate concepts/outcomes? If we’re working on “algebra,” that gives us a ton of latitude, but if we’re working on “integer multiplication,” there’s only so much elaboration to be had—in order to move forward, I assume you have to address topics that for planning and assessment purposes are “currently out of the scope of our class.” Which is a perfectly reasonable practical limitation but maybe also a regrettable one. The only options I see for macro-level differentiation are AP/honors or “skip a grade,” which just seems very lumpy. My intuition is that there’s a lot of opportunity in letting kids sort into more targeted levels in different granularities, like for a couple of days or weeks or a unit or something. I think we do this very well in elementary age reading, because the logistics are relatively straightforward—each kid can just pick a different book! (And umm… learn alone in parallel, which I said above was not great. I need to go think about my life…)
But in other areas (like math) I assume we don’t do that very much because it’s logistically super hard in the first place, and then it makes the problem worse by leading to even greater variation when the kids come back together. It strikes me as a virtuous cycle where the virtue is unfortunately very expensive haha.
In the extreme, this is maybe the boring rebuttal to your floor/ceiling argument—the system today structurally suppresses the development of variation between learners. We simply don’t know how varied learners would be if we allowed them to be assessed variably, either in terms of outcome choice or even just timeline. I think there are convincing practical arguments about how this isn’t really a feasible or high-priority fish to fry, but I do think it’s a valid question.
If you have any thoughts or resources on unit-level differentiation or similar notions, I’d love to move my thoughts here from armchair theorizing to more concrete knowledge. Thanks for engaging!
I have long been an advocate of mastery learning which recognizes the damaging consequences of differences in learning rate that accumulate as groups of students move ahead whether students have learned or not. Some of the original forms of mastery instruction argued that reading was an ideal content presentation mode as it allowed individualization. I don’t see time spent with computers or tablets any more isolating than I see time spent with books. An approach such as Newsela allows the same topic to be addressed at different levels because the same core ideas can be presented at different reading levels. However, learning in some subject areas is heavily sequential (math) and rate of learning is more consequential. Social experiences need not be the same in all content areas even in the elementary grades.
I think Education Realist is right that it is the anti-sociality of Zuck and Gates that trips them up here; they dream of going at their own pace because they think of themselves as better than other people and of reliance on others as dependency.
For everyone else, I think it's incredibly easy to process "learning requires sociality". I was just saying to a student yesterday who wanted to know more about some bodies of 19th Century social theory that I haven't read in a long time that I'd need to re-read some of it in collaboration with colleagues, because I couldn't make much progress all by myself.
Spot on, Dan. Spot on. I nearly shot tea out my nose when I read what they were pivoting to after Summit... AI. They learn nothing ... to your point, the burner is hot and they just can't stop touching it. The universals of good teaching and learning aren't flashy, they don't sell well and therefore doesn't catch the eye of the billionaires. If you were given $100 million to have a positive effect on education, what do you think you would do with it?
Thank you for this great article. Given their influence, I understand the focus on billionaires, but the allure of personalized learning is much, much broader than this exclusive club. I engage with many edtech companies, none of them funded by a tech gazillionaire, and more often than not they pursue the dream of exercises and instruction tailored to individual students. When pressed for a rationale they cite the 2-sigma study of Bloom in the 1980's. But more fundamentally, personalized learning just makes a lot of sense intuitively for people who are not versed in learning science, irrespective of their bank account. This even extends to those working in education (some teachers, many administrators). By rejecting the mirage of personalized learning the billionaires may actually be ahead of the curve ;-)
The number of people trying to reheat the Khan Academy model (this time with AI generated videos! this time with personalized contexts! this time with different gamification features!) is astonishing. Bloom has positively broken the brains of an entire generation of educational technologists.
I believe the reason that this idea has a "stranglehold on billionaires" is that there is tons of research that shows a coach can help a student progress faster than uniform instruction. A student will advance faster at piano, programming, math, or any other discipline with someone helping them when they hit roadblocks.
It's also possible Zuckerberg's initiative failed due to a poor incentive structure. What did the teachers have to gain in implementing the personalized system with enthusiasm? It probably felt exactly as you described it -- as a billionaire reaching down from on high to tell a group of teachers what to do. That's not a likely recipe for success.
> there is tons of research that shows a coach can help a student progress faster than uniform instruction
Before I threw one hundred milli at automated computer multiple choice quizzes I'd want to understand the delta between those quizzes and the kinds of high dosage human tutors that research has found effective.
Fair point! He certainly could have started with a much smaller implementation and iterated it towards success. That type of A/B test would also have been a really interesting educational research contribution.
Would like to hear more about the seemingly fluid back-and-forth between teacher and machine. I have yet to master this transition.
"Students then take time on their own to figure out how many tiles are around several different pools, each time getting automated feedback on their efforts."
Could it be that the technonerds preferred learning alone and assume all other students share their preference? Is all this because some uber-rich guys were socially awkward teens, preferring their high-tech toys to interaction with people?
Amen!! I just wrote something (https://jendycksprout.substack.com/p/the-best-education-is-not-in-our ) along the lines of what you are arguing here and a reader directed me to your piece which has me clapping at my desk. Especially loved how you summarized the problem in this paragraph:
"The idea that computers should personalize instruction, flattening the human differences that learners would much rather see celebrated and developed, maintains an absolute stranglehold on the imagination of the billionaires who fund education technology. Small children will touch a hot stove only once yet billionaires will fund personalized learning initiatives again and again and again and we might wonder why. What accounts for the appeal of this idea? "
This is the sentence I can't get past: "listen to a teacher explain the same material at the same pace in the same way". I don't doubt that many people experienced school that way, or even that there are plenty of classrooms where that is the primary mode of instruction. But the belief that that kind of classroom environment is the best the field of education has to offer is really difficult for me to fathom because if you picked up one book or talked to a single expert it would immediately disabuse you of that belief. I think the relevant question is not "How can we use technology to personalize learning for students?" but "How can we spread 'personalized' learning experiences to more classrooms?" (like the type Dan describes here) which is more of a societal question than a technological one. But tech billionaires only have a hammer and that makes everything a nail.
Another, wonderful example of authentic teaching and learning mathematics. A real problem solved in a community of math curious kids with a teacher who underscores understanding her students' thinking... The public nature of the work and collaboration is so important. Thank you Dan for sharing this.
Amen!! I just wrote something (https://jendycksprout.substack.com/p/the-best-education-is-not-in-our ) along the lines of what you are arguing here and a reader directed me to your piece which has me clapping at my desk. Especially loved how you summarized the problem in this paragraph:
"The idea that computers should personalize instruction, flattening the human differences that learners would much rather see celebrated and developed, maintains an absolute stranglehold on the imagination of the billionaires who fund education technology. Small children will touch a hot stove only once yet billionaires will fund personalized learning initiatives again and again and again and we might wonder why. What accounts for the appeal of this idea? "
I have been reading your blogs for the last year or so and feel so inspired by them as a new math teacher. I believe in group learning and hear my students complain about feeling isolated in personalized classes all the time. This year I am working on designing a gifted education math program in a school that already exists on a mastery learning module. When I bring up the cons of personalized learning to the school leaders, they argue that having such a predesigned curriculum is helpful because not all teachers are trained or experienced enough to have a class that can be successful otherwise. We do live in a place where there is hardly any teacher training and if there is, its not sufficient at all, and yet I do not believe that our only solution is personalized mastery learning courses but I am not sure what to offer instead. Do you have any thoughts?
Thanks for another great piece, Dan - you've been on fire lately. (My lifelong educator mom agrees - she excitedly emailed me this post this morning).
As someone who's also spent time at Summit, I have to point out that you're actually in violent agreement with much of their model from ~2015. They positioned creative group work, led by an excellent teacher like Liz, as the place where the most important learning happened. I actually remember observing a lesson pulled out of your playbook - something to do with creating graphs in Desmos to match an observed phenomenon.
To me, Summit's model accepted as necessary that kids needed "fundamentals" - to acquire a bunch of knowledge and do a bunch of repetitive practice, and the best & most efficient way to get that done was with headphones, laptops, and student independence. I imagine you'd argue that doing that kind work outside of an engaging & meaningful context is ineffective. And the results from Summit Learning may prove you right – maybe any attempt to separate "rote learning" and "meaningful learning" is doomed to failure.
[wahlberg voice] Say hey to your mother for me.
Summit may be catching some strays here. We have hired excellent people from Summit. They were early test sites for our curriculum. (The observed phenomenon graph you mentioned.) If Zuckerberg were investing in the 24 hours of the school week Summit students spend in whole group instruction, or even the full 40, this would be a different conversation. But that 16 hours of locked-into-laptop time seems particularly appealing to billionaire funders in ways that a) are interesting to me, and b) consistently underperform expectations.
I guess the proto-question here which I haven't answered is - when / how much should students work on laptops by themselves, and doing what?
As someone who also spent time with Summit, I'd also suggest that their philosophy is the same as yours. Your example of Liz's sample class is a model of what Summit would want to spend time with. It's separating conceptual understanding from procedural fluency. The online mini-quizzes focusing on process, left me more lesson time to devolop lessons that anticipated, monitored, selected, sequenced and connected student thinking. It was Summit PD that introduced me to that routine. As you know, your 3 Act Tasks were all over the Summit platform.
The model I learned at Summit was not students parked behind a laptop. Even for the 10 pt quiz time, those vilified 16 hours (have you been to classrooms post pandemic? In our school, students don't even vbring paper to class anymore, my class is often the onlyone they need to bring a pencil), we were encouraged to use the data for mini-lessons offline.
I'd suggeat that blaming "out of touch" billionaires is the easy way out of explaining Summit's failure.
What I saw as the fundamental challenge for our community is the same thing that challenges any kind of math reform efforts. It's not like they learned it. Parents, teachers, and administrators alike don't understand this complicated relationship between conceptual knowledge vs. procedural fluency. Students who had been praised for being able to just "do" the procedures (ace the quizzes) didn't see the value in conceptual lessons. It was easy to blame billionaires, rather than opening themselves to a deeper learning experience.
The fact that these happened to be liberal billionaires, and that the curricula happened to include math lessons on identifying racial profiling, questioning the role of folks like Rockefeller and Columbus, made for easy pickings fror our conservative, rural town.
Also, the ambitious administrator and former math teacher who spearheaded the adoption of Summit over the summer resigned the firat day of school, leaving an unsupportive building principal, and a handful of teachers like me who were put up to absorb all of the community's frustration.
As a school, I also don't think we executed the curriculum as intended, and for most classes 100% of the time was behind the laptop. Students would tell me that I was the only one still doing paper and pencil. I have great colleagues, but we were attacked from the beginning, and mostly left to fend for ourselves.
My point here is that I think your analysis on Summits failure throws away the baby with the bathwater. I don't like billionaires any more than you, but the Summit product was good enough for your name, and Desmos to be all over it.
Are you ready to give up on the idea that conceptual understanding and procedural fluency are both important? Sure, lets build teachers focused on assets, not deficits - that seems totally compatible with Summit.
I wouldn't go back to Summit, but the philosophy (if not execution) seems directly in line with what I've been reading from you for nearly 15 years.
Lastly, in answer to your new post about what would you do with $100M, have you considered using it to leverage $100B in annual federal funding to lighten the load on teachers so that weathy techies aren't tempted to plug kids into their products?
" "listen to a teacher explain the same material at the same pace in the same way".
I think this is the disconnect. If you're Zuck or Gates, forced to spend time in a classroom with an inferior teacher with nothing to offer such fine minds as theirs, all they want to do is acquire the information themselves. Education isn't, for them, a social act. They either want all the teacher's attention spent on the "smart" kids or the teacher left for the unworthy, letting the superior kids learn at their own pace, far past the one that mere mortals can manage.
They aren't saying that *now*, of course. I think they are genuinely trying to help. But what they remember from their own classrooms is their own boredom and the wide range of abilities. They never saw school as a social event or interaction.
In that context, it's significant that they are both college dropouts. They didn't want to learn what they didn't want to learn, and there was no value in the social action.
I think this is a pretty likely explanation for the appeal of Laptop Time to billionaire funders.
I think this is a valid critique and an important conversation, but you’re not really engaging with the problem as envisioned by these philanthropic non-educators.
In their view, it is a fundamental problem that all of the students are working on pool border problems at the same time because “that’s what on the schedule.” It’s inefficient and maybe even unjust that kids that already understand the algebra involved have to stay and twiddle their thumbs, and kids that aren’t ready yet have to fake it, because that’s what the logistical constraints of the classroom demand.
Education outsiders observe (I think correctly) that a universal timeline for all learners is really wasteful. We don’t do this because it’s good for learning, we do this because it’s cheap. I think most people would agree that if kids were in much smaller classrooms or received individual tutoring that was targeted to their level, they’d learn a lot more and/or experience much less frustration and grief at being left behind.
The thesis then is that maybe digital technology would make it economical to allow each student to have their own timeline, or at very least, to have some variation based on their ability and prior knowledge. This usually goes poorly because solutions under this thesis tend to ignore all of the social elements of learning that you’re alluding to, because they don’t scale well.
Thanks for the thoughtful response here, Zane. Point of agreement: our current model of large groups of age-graded children accomplishes two goals—one a goal of convenience, as you mention, and the other a goal of socialization. I have not seen a model, especially a computer-based one, that disentangles those goals without great cost.
A point of disagreement, in case you have any follow-up thoughts -
> It’s inefficient and maybe even unjust that kids that already understand the algebra involved have to stay and twiddle their thumbs, and kids that aren’t ready yet have to fake it, because that’s what the logistical constraints of the classroom demand.
My first thought is that every kid is ready to say what they noticed about the pool that Liz flashed on the screen. Every kid is ready to count up the tiles at minimum and many are ready to think about the tile count in more abstract ways. So I want to reject the idea that the floor in math class must necessarily be above the reach of groups of kids even in situations with multi grade levels of learner variability. And I also want to reject the idea that the ceiling is as low as you describe.
I say this as someone who picked up a couple of textbooks of advanced math and skipped from one math class into the next over a two-week holiday break: I knew that math up to a certain level but I didn't understand its depth or connectivity in the least and I would have benefited from someone pushing me to develop my understanding much more.
For example, in this problem, we could ask a student to generalize their solution from an n x n square to a n x m rectangle. We could ask them how an extra layer of tiles would change their solution. We could give them a number of tiles and ask them to tell us the dimensions of the pool, instead of the inverse.
I just want to check what feels like a latent assumption in lots of these conversations, that in a class of 30 kids, only one or two of those kids can get their needs met at max. That isn't the reality in lots of classrooms, even ones with lots of learner variability.
Love this!
Heartily agree that socialization is a worthy goal of classroom learning, at least for me.
I love the idea of varying the difficulty/complexity of the same “problem” via elaboration, complications, etc. This resonates deeply with some lessons from a past life working on world language learning, where I was taught, “change the task, not the text.” I.e. use authentic content that may not be fully comprehensible to all learners, but adapt by not asking them to fully comprehend yet (instead estimate the gist, or enumerate context clues, or recognize key words, etc). Two very differently leveled learners could both benefit from the same content.
I think this is a lot more promising than “every kid on their own, in parallel,” but the skeptic in me wonders how much this relies on the heroics of an individual teacher to surmount structural hurdles. It takes a lot of extra time to elaborate level variations within a topic/learning outcome/problem. And then at the end of the day/unit will we still assess students uniformly? I’m here for learning for its own sake, even if it never shows up on a test, but when push comes to shove anything that’s not on the test tends to get cut first, right?
Then I think about how variation might compound over months and years. In terms of the granularity of the variation, you’re looking at an individual problem/lesson. Which is awesome! But at some point doesn’t that naturally mean that some learners ought to be on totally separate concepts/outcomes? If we’re working on “algebra,” that gives us a ton of latitude, but if we’re working on “integer multiplication,” there’s only so much elaboration to be had—in order to move forward, I assume you have to address topics that for planning and assessment purposes are “currently out of the scope of our class.” Which is a perfectly reasonable practical limitation but maybe also a regrettable one. The only options I see for macro-level differentiation are AP/honors or “skip a grade,” which just seems very lumpy. My intuition is that there’s a lot of opportunity in letting kids sort into more targeted levels in different granularities, like for a couple of days or weeks or a unit or something. I think we do this very well in elementary age reading, because the logistics are relatively straightforward—each kid can just pick a different book! (And umm… learn alone in parallel, which I said above was not great. I need to go think about my life…)
But in other areas (like math) I assume we don’t do that very much because it’s logistically super hard in the first place, and then it makes the problem worse by leading to even greater variation when the kids come back together. It strikes me as a virtuous cycle where the virtue is unfortunately very expensive haha.
In the extreme, this is maybe the boring rebuttal to your floor/ceiling argument—the system today structurally suppresses the development of variation between learners. We simply don’t know how varied learners would be if we allowed them to be assessed variably, either in terms of outcome choice or even just timeline. I think there are convincing practical arguments about how this isn’t really a feasible or high-priority fish to fry, but I do think it’s a valid question.
If you have any thoughts or resources on unit-level differentiation or similar notions, I’d love to move my thoughts here from armchair theorizing to more concrete knowledge. Thanks for engaging!
I have long been an advocate of mastery learning which recognizes the damaging consequences of differences in learning rate that accumulate as groups of students move ahead whether students have learned or not. Some of the original forms of mastery instruction argued that reading was an ideal content presentation mode as it allowed individualization. I don’t see time spent with computers or tablets any more isolating than I see time spent with books. An approach such as Newsela allows the same topic to be addressed at different levels because the same core ideas can be presented at different reading levels. However, learning in some subject areas is heavily sequential (math) and rate of learning is more consequential. Social experiences need not be the same in all content areas even in the elementary grades.
> I don’t see time spent with computers or tablets any more isolating than I see time spent with books.
I agree. Neither is adequate for the core of most students' learning.
I think Education Realist is right that it is the anti-sociality of Zuck and Gates that trips them up here; they dream of going at their own pace because they think of themselves as better than other people and of reliance on others as dependency.
For everyone else, I think it's incredibly easy to process "learning requires sociality". I was just saying to a student yesterday who wanted to know more about some bodies of 19th Century social theory that I haven't read in a long time that I'd need to re-read some of it in collaboration with colleagues, because I couldn't make much progress all by myself.
Spot on, Dan. Spot on. I nearly shot tea out my nose when I read what they were pivoting to after Summit... AI. They learn nothing ... to your point, the burner is hot and they just can't stop touching it. The universals of good teaching and learning aren't flashy, they don't sell well and therefore doesn't catch the eye of the billionaires. If you were given $100 million to have a positive effect on education, what do you think you would do with it?
Thank you for this great article. Given their influence, I understand the focus on billionaires, but the allure of personalized learning is much, much broader than this exclusive club. I engage with many edtech companies, none of them funded by a tech gazillionaire, and more often than not they pursue the dream of exercises and instruction tailored to individual students. When pressed for a rationale they cite the 2-sigma study of Bloom in the 1980's. But more fundamentally, personalized learning just makes a lot of sense intuitively for people who are not versed in learning science, irrespective of their bank account. This even extends to those working in education (some teachers, many administrators). By rejecting the mirage of personalized learning the billionaires may actually be ahead of the curve ;-)
The number of people trying to reheat the Khan Academy model (this time with AI generated videos! this time with personalized contexts! this time with different gamification features!) is astonishing. Bloom has positively broken the brains of an entire generation of educational technologists.
Or, they are people like me who got to experience mastery learning in an analog form and both and enjoyed and greatly benefitted from it.
I believe the reason that this idea has a "stranglehold on billionaires" is that there is tons of research that shows a coach can help a student progress faster than uniform instruction. A student will advance faster at piano, programming, math, or any other discipline with someone helping them when they hit roadblocks.
It's also possible Zuckerberg's initiative failed due to a poor incentive structure. What did the teachers have to gain in implementing the personalized system with enthusiasm? It probably felt exactly as you described it -- as a billionaire reaching down from on high to tell a group of teachers what to do. That's not a likely recipe for success.
> there is tons of research that shows a coach can help a student progress faster than uniform instruction
Before I threw one hundred milli at automated computer multiple choice quizzes I'd want to understand the delta between those quizzes and the kinds of high dosage human tutors that research has found effective.
Fair point! He certainly could have started with a much smaller implementation and iterated it towards success. That type of A/B test would also have been a really interesting educational research contribution.
Would like to hear more about the seemingly fluid back-and-forth between teacher and machine. I have yet to master this transition.
"Students then take time on their own to figure out how many tiles are around several different pools, each time getting automated feedback on their efforts."
Love your content and approach! I write my own education newsletter. It is nice to discover kindred spirits on Substack. Keep up the amazing work. 👍👍
Could it be that the technonerds preferred learning alone and assume all other students share their preference? Is all this because some uber-rich guys were socially awkward teens, preferring their high-tech toys to interaction with people?
Amen!! I just wrote something (https://jendycksprout.substack.com/p/the-best-education-is-not-in-our ) along the lines of what you are arguing here and a reader directed me to your piece which has me clapping at my desk. Especially loved how you summarized the problem in this paragraph:
"The idea that computers should personalize instruction, flattening the human differences that learners would much rather see celebrated and developed, maintains an absolute stranglehold on the imagination of the billionaires who fund education technology. Small children will touch a hot stove only once yet billionaires will fund personalized learning initiatives again and again and again and we might wonder why. What accounts for the appeal of this idea? "