As a retired teacher, I feel I can attest to who benefits from a technology driven education. The teachers and the administration are under pressure to constantly provide progress data. When students' work is primarily on their computers, that can be achieved. It doesn't benefit the student but it reduces the workload of faculty. Education has devolved into statistical data and is no longer about creating a classroom of lifelong learners. It resembles an assembly line with machines creating identical models, only these are people, not automobiles.
I concur as a retired educator who was part of an ed-tech implementation that trained
our team, as content area specialists, to serve potential dropouts in a relationally supportive small-school context. We had a significant, positive impact over eight years. The tech-based courses were hit or miss (we ended up creating our own digital options). What WORKED was the teacher facilitation, offline support, required offline demonstrations of understanding, tight limits on assessment takes and retakes, flexibility and a seat-time waiver, cross-curricular service projects in our community…
Alas, powers at the district level determined they could get more bang for their buck by closing our teacher-led program and shifting the online courses back to the local high schools where these same types of students sit in a computer lab with a non-certified para-professional, no checks on preparation before assessments, unlimited “retakes” (notable that all assessment answers are available via Discord & Reddit) —> ever higher graduation rates.
The article is correct about one point: “no product-specific causal proof of harm” can serve as a convenient corporate standard when products evolve faster than research cycles. The problem arises in what it does next. It uses that procedural point to justify a substantive leap: from a specific type of social media risk (appearance manipulation, social comparison, engagement incentives) to a broad, catch-all category called “EdTech,” and then it treats that category as if it’s one exposure with a single mechanism.
“EdTech” is not a single entity. It encompasses calculators, audiobooks, text-to-speech, classroom projectors, learning management systems, adaptive tutoring, and high-stakes online testing. Some of these ease the learning process; some replace learning; some are neutral; and some can be harmful depending on age, usage, and what they substitute. If calculators are considered “technology in classrooms” (which they are), then the argument cannot simply be that “technology is damaging.” The real argument is that certain technologies, when used at developmental stages, in specific ways, can produce trade-offs. This is a different claim—one that needs definition and clarity.
The Utah NAEP story is presented as a “structural break” beginning around the introduction of a computerised assessment system. Even if the trendline description is accurate, the causal suggestion is not justified. A correlation between a statewide testing shift and later NAEP slopes does not identify the mechanism, and it does not distinguish school technology from the many other simultaneous changes that plausibly affect reading and maths: smartphones and social media saturation outside school, changes in reading habits, curriculum and standards adjustments, demographic change, teacher labour market dynamics, pandemic aftershocks beyond a single test year, and simple regression-to-the-mean after decades of improvement. If the claim is “classroom devices and platforms caused the decline,” you need evidence of exposure (time-on-device), substitution (what instruction was displaced), and a plausible pathway (attention fragmentation, reduced reading volume, reduced handwriting, weaker teacher-led instruction), ideally tested across multiple jurisdictions with appropriate controls. Without that, “converging evidence” is a rhetorical label applied to an under-specified hypothesis.
The ethics argument is also being used too broadly. It’s true that you often cannot run long, high-risk randomised trials in children. But that doesn’t justify collapsing categories and regulating by insinuation. Ethical limits should push policymakers toward better observational designs, clearer definitions, transparent procurement standards, and staged rollouts with measurable outcomes—not toward treating “digital” as the hazard.
If the author wants a defensible policy stance, it’s available without overreach. Shift the responsibility to vendors and districts to demonstrate durable, replicable benefits for instructional platforms before scaling. Require evaluations that measure not only short-term engagement but also learning transfer and retention. Limit entertainment-style features in student-facing products. Age-gate and minimise devices in early primary education where foundational skills are being automated internally. Protect time for reading, handwriting, and sustained attention. Maintain proven “utility tech” (calculators at the appropriate stage, accessibility tools like text-to-speech, assistive communication) precisely because the goal is learning, not a war on tools.
The article’s strongest point is opportunity cost. If districts spend heavily on platforms that don’t outperform well-implemented analogue instruction, that’s a serious policy failure. But “we wasted money” is not the same claim as “technology is damaging,” and it shouldn’t be conflated into one. A serious debate about schooling needs to discuss specifics: which tool, which age group, which dosage, what is being replaced, how it is measured, and compared to what alternative.
While I agree with your points generally, I think politically the only way to turn this ship around is to use the more harmful examples to scare constituents and lawmakers into giving a f***. School districts should be scared out of their minds that little Johnny accidentally sees something sexually graphic on his school-issued ipad, but instead there are literally no repercussions for these offenses or damage. The pendulum needs to swing all the way back before it could settle into some sort of balance. We need laws to protect kids from harmful content while they are at school, the same way we already have laws and systems in place to protect them from harmful behaviors and people.
This comment is on point. EdTech is way too broad a category for the argument made. I also worry about broad stroke 'bans' without causal evidence. I do think that schools can be thoughtful about what is beneficial and harmful and put measures in place especially as they apply to different ages or groups or even individual students that address technology that changes faster than research can study.
There also needs to be a gradual transfer of responsibility. The role of schools is too nuture and teach. If we remove that responsibility entirely as it relates to technology (something that students have access to outside of school walls) where will they learn in a safe environment? This is not an all or nothing topic. All technologies in schools are arrival technologies (Reich and Dukes) and as such we have a responsibility to help teach digital literacy through education and use.
While I appreciate the due diligence you and Kevin are demanding and agree that all arguments should be as solid as possible, I think you are missing the point.
I don't believe that "all or nothing" is the argument being made here. The issue is awareness. The apparent cause for alarm is due to the fact that it is already too late. Instead of nitpicking whether or not the "brush is too big", pay attention to the heart of what is being shared. As an educator for over 20 years, I can attest to "the kids are not alright" and tech dependence, yes, at home and at school, has clear implications for that diagnosis. What Horvath is arguing here is that we need to be intentional about what we expose our youngest citizens to, and we can't trust the creators of tech, ed, or otherwise, to be honest with us. Therefore, we must be vigilant. This isn't about whether calculators harm our students. This is about who we can trust in the realm of EdTech. Our children are too important not to take this seriously.
I don’t disagree with the urgency or the trust issue. Vendors don’t get to be the sole arbiters of what’s “safe” for children.
But I think this thread is still missing the upstream point.
The way I see it, we keep trying to locate “the villain” within a single downstream object (filters, apps, devices), when the deeper problem is the ecology that completes the child’s wiring. Cognition isn’t like haemoglobin — it’s a high Degrees-of-Constraint system that emerges inside stacked environmental constraints.
When that environment becomes non-negotiable, outcomes funnel, and prevalence can rise without the genome changing.
That’s why I’m wary of arguing purely at the “EdTech” level, especially when “EdTech” isn’t defined. Calculators, text-to-speech, locked-down assessments, open browsers, gamified engagement apps — these are not a single exposure with a single mechanism. If we don’t separate categories, “vigilance” becomes blunt policy theatre.
Reclaiming childhood means reclaiming the conditions the experience-expectant brain is waiting for: regulation, movement, co-regulation, language rhythm, touch, sleep, nourishment, safe stress and repair.
And in 2026, we also have to admit that childcare now occupies a large share of that attachment/regulation window for many children, so this can’t be treated as “just” a school IT issue.
So yes: enforce safeguarding on school devices as non-negotiable. But if we want genuine reversal rather than whack-a-mole, we have to look upstream at the attachment and regulation ecology that’s shaping children long before any specific platform does.
From what I have heard Zuckerberg and others in silicon valley send their children to traditional, tech free schools. If so, that is proof enough that he does not see any advantage in computors/screens as learning devices. At least not enough to entrust his childrens education with it. When harm can be proven it is already too late to rectify the damage of an entire generation or more.
I used to work for an educational publisher and discovered over the years that the field of education is exceedingly prone to fads. Every year a brand new technology or teaching technique promised the moon. I got pretty fed up after a while and tuned out all the latest innovations. I would routinely visit schools located in districts ranging from the richest to the poorest. One of them was a science class in a very small middle school in Brooklyn that had practically nothing in the way of equipment. I was shocked at how bare the classroom was. Yet the kids were extremely curious and enthusiastic and clearly loved their teacher.
This reminds me of the 2013 talk Malcolm Gladwell gave at UPenn about the burden of proof to connect CTE and concussions suffered playing college football. When there is no agreed upon level of proof at which we will take action the result is an endless cycle demanding more evidence of harm. Whether that is inertia, learned helplessness, or simply that we desperately want to believe it only affects other people, far from ourselves it is hard to tell.
Absence of evidence is not evidence of absence. It's a basic point that so many people choose to ignore when they want to advance a big change in anything.
Second, they warn against these sorts of correlations because there are so many experimental and small-scale studies that show positive effects of technological applications on student performance in various subject areas, such as literacy, mathematics, and science. See:
I'll also quote John Hattie here: “The message... is not that technology has not made a difference. It has, and that is what an effect of 0.3 to 0.5 signifies....If there is to be a third edition of Visual Learning, there would not be a chapter on technology influences, as it is time to think of technology like curricula, books, and other resources, and focus more on the implementation effects when including technologies and teaching methods, the when and how within the teaching cycle, the dosage and the fidelity of various apps, and so on….”(p.408, Visible Learning)
Please explain why all of these researchers, and the researchers who peer-reviewed their work before publication, are wrong.
"No evidence of harm" is also used by "Gender Affirming Care" providing cross/wrong-sex hormones and removing healthy body parts of our children in the name of gender ideology.
You stated, "When billions of dollars and millions of children are involved, the burden of proof should rest on demonstrating clear, durable, replicable benefit — not on proving harm after the fact.
Caution is not fear, and restraint is not regression. They are marks of a society that prioritizes children over products." This statement applies to the harm happening to our kids with a medicalization pathway that prioritizes cosmetic, irreversible medical interventions. Puberty blockers halt natural child development. "Gender medicine" will not examine the clear evidence of harm that is being exposed and will not admit that children cannot consent to altering their natural development.
Our society, focused on gender identity, has eliminated protective safeguards for our children and deflected in a similar manner as those who run social media platforms. I appreciate your article. Our advocacy is linked in the goal of protecting children and restoring their childhood free of the harms of social media and medical interventions upon their bodies. Both have long lasting and devastating effects.
These Utah stats are alarming. What do the data from other states show? Moreover, I would love to know how the full integration of tech into education has affected outcomes in other countries. Does anyone know of additional data?
As a retired teacher, I feel I can attest to who benefits from a technology driven education. The teachers and the administration are under pressure to constantly provide progress data. When students' work is primarily on their computers, that can be achieved. It doesn't benefit the student but it reduces the workload of faculty. Education has devolved into statistical data and is no longer about creating a classroom of lifelong learners. It resembles an assembly line with machines creating identical models, only these are people, not automobiles.
I concur as a retired educator who was part of an ed-tech implementation that trained
our team, as content area specialists, to serve potential dropouts in a relationally supportive small-school context. We had a significant, positive impact over eight years. The tech-based courses were hit or miss (we ended up creating our own digital options). What WORKED was the teacher facilitation, offline support, required offline demonstrations of understanding, tight limits on assessment takes and retakes, flexibility and a seat-time waiver, cross-curricular service projects in our community…
Alas, powers at the district level determined they could get more bang for their buck by closing our teacher-led program and shifting the online courses back to the local high schools where these same types of students sit in a computer lab with a non-certified para-professional, no checks on preparation before assessments, unlimited “retakes” (notable that all assessment answers are available via Discord & Reddit) —> ever higher graduation rates.
The article is correct about one point: “no product-specific causal proof of harm” can serve as a convenient corporate standard when products evolve faster than research cycles. The problem arises in what it does next. It uses that procedural point to justify a substantive leap: from a specific type of social media risk (appearance manipulation, social comparison, engagement incentives) to a broad, catch-all category called “EdTech,” and then it treats that category as if it’s one exposure with a single mechanism.
“EdTech” is not a single entity. It encompasses calculators, audiobooks, text-to-speech, classroom projectors, learning management systems, adaptive tutoring, and high-stakes online testing. Some of these ease the learning process; some replace learning; some are neutral; and some can be harmful depending on age, usage, and what they substitute. If calculators are considered “technology in classrooms” (which they are), then the argument cannot simply be that “technology is damaging.” The real argument is that certain technologies, when used at developmental stages, in specific ways, can produce trade-offs. This is a different claim—one that needs definition and clarity.
The Utah NAEP story is presented as a “structural break” beginning around the introduction of a computerised assessment system. Even if the trendline description is accurate, the causal suggestion is not justified. A correlation between a statewide testing shift and later NAEP slopes does not identify the mechanism, and it does not distinguish school technology from the many other simultaneous changes that plausibly affect reading and maths: smartphones and social media saturation outside school, changes in reading habits, curriculum and standards adjustments, demographic change, teacher labour market dynamics, pandemic aftershocks beyond a single test year, and simple regression-to-the-mean after decades of improvement. If the claim is “classroom devices and platforms caused the decline,” you need evidence of exposure (time-on-device), substitution (what instruction was displaced), and a plausible pathway (attention fragmentation, reduced reading volume, reduced handwriting, weaker teacher-led instruction), ideally tested across multiple jurisdictions with appropriate controls. Without that, “converging evidence” is a rhetorical label applied to an under-specified hypothesis.
The ethics argument is also being used too broadly. It’s true that you often cannot run long, high-risk randomised trials in children. But that doesn’t justify collapsing categories and regulating by insinuation. Ethical limits should push policymakers toward better observational designs, clearer definitions, transparent procurement standards, and staged rollouts with measurable outcomes—not toward treating “digital” as the hazard.
If the author wants a defensible policy stance, it’s available without overreach. Shift the responsibility to vendors and districts to demonstrate durable, replicable benefits for instructional platforms before scaling. Require evaluations that measure not only short-term engagement but also learning transfer and retention. Limit entertainment-style features in student-facing products. Age-gate and minimise devices in early primary education where foundational skills are being automated internally. Protect time for reading, handwriting, and sustained attention. Maintain proven “utility tech” (calculators at the appropriate stage, accessibility tools like text-to-speech, assistive communication) precisely because the goal is learning, not a war on tools.
The article’s strongest point is opportunity cost. If districts spend heavily on platforms that don’t outperform well-implemented analogue instruction, that’s a serious policy failure. But “we wasted money” is not the same claim as “technology is damaging,” and it shouldn’t be conflated into one. A serious debate about schooling needs to discuss specifics: which tool, which age group, which dosage, what is being replaced, how it is measured, and compared to what alternative.
While I agree with your points generally, I think politically the only way to turn this ship around is to use the more harmful examples to scare constituents and lawmakers into giving a f***. School districts should be scared out of their minds that little Johnny accidentally sees something sexually graphic on his school-issued ipad, but instead there are literally no repercussions for these offenses or damage. The pendulum needs to swing all the way back before it could settle into some sort of balance. We need laws to protect kids from harmful content while they are at school, the same way we already have laws and systems in place to protect them from harmful behaviors and people.
This comment is on point. EdTech is way too broad a category for the argument made. I also worry about broad stroke 'bans' without causal evidence. I do think that schools can be thoughtful about what is beneficial and harmful and put measures in place especially as they apply to different ages or groups or even individual students that address technology that changes faster than research can study.
There also needs to be a gradual transfer of responsibility. The role of schools is too nuture and teach. If we remove that responsibility entirely as it relates to technology (something that students have access to outside of school walls) where will they learn in a safe environment? This is not an all or nothing topic. All technologies in schools are arrival technologies (Reich and Dukes) and as such we have a responsibility to help teach digital literacy through education and use.
While I appreciate the due diligence you and Kevin are demanding and agree that all arguments should be as solid as possible, I think you are missing the point.
I don't believe that "all or nothing" is the argument being made here. The issue is awareness. The apparent cause for alarm is due to the fact that it is already too late. Instead of nitpicking whether or not the "brush is too big", pay attention to the heart of what is being shared. As an educator for over 20 years, I can attest to "the kids are not alright" and tech dependence, yes, at home and at school, has clear implications for that diagnosis. What Horvath is arguing here is that we need to be intentional about what we expose our youngest citizens to, and we can't trust the creators of tech, ed, or otherwise, to be honest with us. Therefore, we must be vigilant. This isn't about whether calculators harm our students. This is about who we can trust in the realm of EdTech. Our children are too important not to take this seriously.
I don’t disagree with the urgency or the trust issue. Vendors don’t get to be the sole arbiters of what’s “safe” for children.
But I think this thread is still missing the upstream point.
The way I see it, we keep trying to locate “the villain” within a single downstream object (filters, apps, devices), when the deeper problem is the ecology that completes the child’s wiring. Cognition isn’t like haemoglobin — it’s a high Degrees-of-Constraint system that emerges inside stacked environmental constraints.
When that environment becomes non-negotiable, outcomes funnel, and prevalence can rise without the genome changing.
That’s why I’m wary of arguing purely at the “EdTech” level, especially when “EdTech” isn’t defined. Calculators, text-to-speech, locked-down assessments, open browsers, gamified engagement apps — these are not a single exposure with a single mechanism. If we don’t separate categories, “vigilance” becomes blunt policy theatre.
Reclaiming childhood means reclaiming the conditions the experience-expectant brain is waiting for: regulation, movement, co-regulation, language rhythm, touch, sleep, nourishment, safe stress and repair.
And in 2026, we also have to admit that childcare now occupies a large share of that attachment/regulation window for many children, so this can’t be treated as “just” a school IT issue.
So yes: enforce safeguarding on school devices as non-negotiable. But if we want genuine reversal rather than whack-a-mole, we have to look upstream at the attachment and regulation ecology that’s shaping children long before any specific platform does.
From what I have heard Zuckerberg and others in silicon valley send their children to traditional, tech free schools. If so, that is proof enough that he does not see any advantage in computors/screens as learning devices. At least not enough to entrust his childrens education with it. When harm can be proven it is already too late to rectify the damage of an entire generation or more.
I used to work for an educational publisher and discovered over the years that the field of education is exceedingly prone to fads. Every year a brand new technology or teaching technique promised the moon. I got pretty fed up after a while and tuned out all the latest innovations. I would routinely visit schools located in districts ranging from the richest to the poorest. One of them was a science class in a very small middle school in Brooklyn that had practically nothing in the way of equipment. I was shocked at how bare the classroom was. Yet the kids were extremely curious and enthusiastic and clearly loved their teacher.
This reminds me of the 2013 talk Malcolm Gladwell gave at UPenn about the burden of proof to connect CTE and concussions suffered playing college football. When there is no agreed upon level of proof at which we will take action the result is an endless cycle demanding more evidence of harm. Whether that is inertia, learned helplessness, or simply that we desperately want to believe it only affects other people, far from ourselves it is hard to tell.
Absence of evidence is not evidence of absence. It's a basic point that so many people choose to ignore when they want to advance a big change in anything.
You're continuing to ignore that there is evidence of benefit.
First, researchers warn against exactly what you're doing in this post: https://www.sciencedirect.com/science/article/pii/S0360131521002384#:~:text=Evidence%20of%20the%20impact%20of,2015;%20Spiezia%2C%202010).
Second, they warn against these sorts of correlations because there are so many experimental and small-scale studies that show positive effects of technological applications on student performance in various subject areas, such as literacy, mathematics, and science. See:
https://www.sciencedirect.com/science/article/abs/pii/S0360131516302172
https://www.sciencedirect.com/science/article/abs/pii/S1747938X12000401
https://www.sciencedirect.com/science/article/abs/pii/S1747938X13000031
https://www.sciencedirect.com/science/article/pii/S0360131515300804
And most recently: https://journals.sagepub.com/doi/abs/10.3102/00346543241261073?ai=1gvoi&mi=3ricys&af=R
I'll also quote John Hattie here: “The message... is not that technology has not made a difference. It has, and that is what an effect of 0.3 to 0.5 signifies....If there is to be a third edition of Visual Learning, there would not be a chapter on technology influences, as it is time to think of technology like curricula, books, and other resources, and focus more on the implementation effects when including technologies and teaching methods, the when and how within the teaching cycle, the dosage and the fidelity of various apps, and so on….”(p.408, Visible Learning)
Please explain why all of these researchers, and the researchers who peer-reviewed their work before publication, are wrong.
"No evidence of harm" is also used by "Gender Affirming Care" providing cross/wrong-sex hormones and removing healthy body parts of our children in the name of gender ideology.
You stated, "When billions of dollars and millions of children are involved, the burden of proof should rest on demonstrating clear, durable, replicable benefit — not on proving harm after the fact.
Caution is not fear, and restraint is not regression. They are marks of a society that prioritizes children over products." This statement applies to the harm happening to our kids with a medicalization pathway that prioritizes cosmetic, irreversible medical interventions. Puberty blockers halt natural child development. "Gender medicine" will not examine the clear evidence of harm that is being exposed and will not admit that children cannot consent to altering their natural development.
Our society, focused on gender identity, has eliminated protective safeguards for our children and deflected in a similar manner as those who run social media platforms. I appreciate your article. Our advocacy is linked in the goal of protecting children and restoring their childhood free of the harms of social media and medical interventions upon their bodies. Both have long lasting and devastating effects.
Thank you for this! So, so important!
Thank you for writing this and fighting this fight for our kids
These Utah stats are alarming. What do the data from other states show? Moreover, I would love to know how the full integration of tech into education has affected outcomes in other countries. Does anyone know of additional data?
Horvath details all of these things in his book Digital Delusion.
Well done. Thank you.