Jonathan Haidt's book The Anxious Generation made a big splash in education circles last year. It elaborates the thesis that social media is harming people's mental health, especially that of teenagers and young adults. Haidt has compiled a lot of different evidence that he says supports this conclusion, but some of his academic colleagues aren't buying it.
To briefly summarize Haidt’s position, there is strong correlational evidence based on two co-occurring trends: The specific timeframe from 2010-2015 showed a dramatic emergency of new mental health problems on a wide range of measures for teenagers, especially younger adolescents ages 10-14. At the same time, smartphones made social media apps continuously available to people in this age bracket. Haidt argues that it was the specific combination of social media apps and the internet in one’s pocket that made the situation so harmful, which helps him to explain why this specific 5-year period was the point at which mental health issues skyrocketed. Social media apps were available before that time, but teenagers’ internet access was more restricted — perhaps using a shared family computer instead of a personal device, and definitely not during school hours.
Haidt notes that there was no corresponding dramatic rise in mental health issues for older or younger cohorts, which suggests that the effect was tied to people going through a developmental sensitive period. The early teen years are known to be crucial for social development, and Haidt lays out several logical ways in which social media apps can interfere with these tasks. For example, they are disembodied, can be anonymous, and can involve curating personal content for a large public audience. Social pressure is especially strong for girls, and they show the worst mental health outcomes. They also have more focus on physical appearance, more relational forms of aggression, more willingness to share emotions, and more vulnerability to online sexual predators (mainly older men). For boys the main risks are video games and pornography, which can interfere with real world relationships but are potentially less damaging to self-concept than social media are for girls, and boys do show correspondingly less increase in mental health symptoms (although still a concerning amount). For both genders, Haidt argues that the 4 major harms to teenagers are (1) time spent online at the expense of real-world learning and relationships, (2) significant sleep deprivation as time spent online competes with rest, (3) attention fragmentation caused by badges and notifications, and (4) addiction to the small hit of dopamine that is released every time you see a bit of personalized content online.
Haidt also spends about half of the book arguing that adolescents need more opportunity for real-life experiences, better play spaces, less supervision, and more independence at an earlier age. He says that unstructured play is the natural way to support adolescents’ social development. Those suggestions seemed less strongly supported by research. For example, Haidt himself acknowledges that the trend toward “safetyism” has been building for decades, far before the advent of either smartphones or social media. Because of that, these ideas seem more ideological or like old-guy griping: “We had much more dangerous metal playgrounds in my day. My best friend got caught on one and ripped out his spleen. He ended up just fine!” (True story, by the way). One plausible way in which over-protecting kids might be contributing to the problem is by making online communities their only opportunity to socialize, because public spaces have become less big-kid-friendly. My daughter said that she and her friends basically hung out at each other’s houses in high school, because there was no place else open to them – a group of teens in a park aroused suspicion, malls have largely been replaced by big-box stores, and they were expected to be off school property shortly after class was done. When public spaces are closed to children, and parents may be reported for neglect if they let younger children roam as freely as they did during their own childhoods, online socialization can feel like the safe option. On balance I tend to agree with Haidt’s gripes in this area, and think that adolescents might do better and use online media less if they were given more freedom than they have. But then, I’m an old guy too – and what we need in this area is data, not opinions.
Haidt’s detractors have one main critique, which is that correlation is not causation. This is such a standard attack in academic punching matches that it almost goes without saying, but on the other hand it is so widely used because it is so effective. To the critic who demands only evidence from randomized controlled trials (RCTs), the evidence is inconsistent. Several meta-analyses have been attempted, and it appears that what you get depends on how you set your criteria for study quality. For example, studies that tested a 1-week social media “fast” fail to show improvements in mental health, while some studies of 2-week or longer abstinence do demonstrate benefits. Studies of this type are also hard to conduct because of collective action problems, which Haidt ably discusses in the book. For instance, children feel left out when all their friends are on social media, children pressure their parents to use phones, and “cheating” by peeking at social media accounts is probably common with some participants even resorting to sneaky tactics such as using a parent’s computer behind their back. Besides all of these examples where investigators lack appropriate experimental control, it can be hard to get people to agree to participate in such studies, and those who do might be outliers compared to the general population. Cluster-randomizing schools so that some entire communities stop using their phones (or social media apps in particular) are likely the best method, but such studies are tricky and expensive to conduct. And then there are all the potential variations: one week, two, or longer? Some subset of specific apps, or everything digital? Phones in pockets or locked up at the school office? Use prohibited during breaks and passing periods,Mor only during class? How to accommodate online schoolwork, family entertainment, or parents’ desire to check in with their kids throughout the day? All of these design features are likely to have an impact on the experimental results of any given study. And all current studies are necessarily conducted in an environment where phone-based social media use is already the norm for teenagers. To truly test Haidt’s central hypothesis about the timeframe when so many of teenagers’ mental health problems first appeared, it would be necessary to have done all of this research back in 2012.
I found the most compelling aspect of Haidt’s argument to be the sense of exploitation that taints companies’ relationships with their social-media customers, particularly teens. He reports a personal interview with Mark Zuckerberg about companies’ failure to enforce age rules for account holders in any effective way (“we’re working on that” was the only answer). In the meantime, companies invest vast resources in what Haidt calls a “race to the bottom of the brain stem,” optimizing features of systems to monopolize attention and increase total time spent using the . Key strategies for the most successful social media are personalization (i.e., users post their own content), approval ratings (by you and of you), using AI to optimize each user’s feed based on observed patterns of behavior rather than the user’s consciously chosen preferences, and reducing friction to keep users engaged for long periods of time, e.g. by using “breaking news” alerts and an endless scroll. In 2025, social AI chatbots are adding a new dimension of personalization and therefore of risk. Haidt blames companies for knowingly making their algorithms addictive to users, and presents evidence from internal Meta (Facebook/Instagram) company memos showing that the company deliberately capitalized on its underage users’ mental immaturity.
Some of Haidt’s suggestions for government and industry seem like common sense solutions, for instance to assert that companies have an affirmative duty of care for minors that would require them to consider their users’ best interests instead of just their bottom lines. He also suggests raising the minimum account registration age from 13 to 16 (which was the original proposed Federal rule in 2012), and to facilitate age verification – e.g., by using blockchain ID or by registering a device to indicate that it belongs to a minor. Importantly, both of these solutions allow for age rule enforcement without compromising online privacy, as opposed to other solutions like requiring someone to upload a photo of their driver’s license. I was particularly impressed by the simplicity of these solutions compared to the more commonly discussed idea of content moderation, which smacks of censorship and makes people worry about particular political or other viewpoints being suppressed. Some type of device registration would be much more akin to keeping teens out of the liquor store: Sorry, your device can’t come in. Check back when you are 18. Could teens use a friend’s computer? Yes, just like they can get a fake ID. But it would be so much stronger as a form of protection than the current “click here to verify you are 13” box.
As we move from global to local levels of policy, Haidt’s solutions become a bigger lift. He recommends completely phone-free schools: not even in a pocket, where some research shows it has a distracting effect. Forbidding use during class time only doesn’t have the same benefits, yet most students and parents will push for this. At my younger daughter’s high school, the principal read Anxious Generation last summer and started the year at 100% phone-free rules, with harsh consequences including detention on the 2nd offense and in-school suspension for the 3rd. The minute the rules were announced, parents and teachers started lobbying for exceptions. And halfway through the school year, the policy-as-enforced is a mere shadow of the policy-as-designed, well below Haidt’s suggested minimum criteria for effectiveness. And as for parents, even Haidt is pessimistic about their ability to make progress with an individual child’s phone use. Instead, he suggests banding together with like-minded peers to work toward phone-free communities. And of course to break our own addiction to the tiny screens in our pockets, which is a lot easier said than done.
My final impression of The Anxious Generation is that it’s an important book, calling attention to a specific aspect of the technology-use problem – social media – and specific age bracket – 10 to 14. This level of specificity makes it easier to take action and test results. I was also relatively convinced by the arguments for a connection between social app use and mental health problems, despite their largely correlational nature. Unfortunately, solutions remain difficult, and it will be hard to produce research that’s convincing enough to prompt major policy changes. That doesn’t mean we shouldn’t try, because the mental health burden for young people in 2025 is staggering.
Comments
Post a Comment