Haidt's The Anxious Generation — A Steelman
The strongest, fairest version of the argument — so you can agree with what's right in it before you disagree with the rest.
The weak version everyone argues with is "phones are bad." Ignore it. The strong version is a developmental-displacement argument with a specific mechanism, specific evidence, and a specific reason individual effort can't fix it.
The core claim. Childhood underwent what Haidt calls a "Great Rewiring" between roughly 2010 and 2015. A play-based childhood — unstructured, physical, risky, in-person — was rapidly swapped for a phone-based childhood. That timing matters because adolescence is a neurologically sensitive window: the brain's reward and emotional systems mature faster than the executive-control system meant to govern them, so whatever fills those years gets wired in. His framing line is the sharpest thing in the book: we have become overprotective in the real world and underprotective online — we stopped letting kids roam the neighborhood and started letting them roam the internet.
The mechanism. Four foundational harms: social deprivation, sleep deprivation, attention fragmentation, and addiction. Crucially, these aren't mainly about bad content. Even a perfectly clean feed displaces the hours that used to go to sleep, face-to-face friendship, movement, and boredom. The harm is the opportunity cost as much as the exposure. Two pathways lead to the same place: girls primarily via social media (comparison and relational aggression, with 11–13 the most dangerous window), boys via gaming, porn, and withdrawal into virtual worlds — both ending in anomie, a felt sense that life is meaningless.
Why he says it's not a panic (his best defensive move). "Social media is not like sugar." Sugar harms the consumer; social media has group-level effects — it changes the social environment for everyone, which makes it a collective-action trap. No single parent can opt their kid out without isolating that kid socially. This is the move that (a) justifies policy rather than leaving it to families, and (b) pre-empts "just parent better." It's the load-bearing argument for bans and phone-free schools.
The evidence he leans on hardest. The timing (the hockey-stick rise in teen depression, anxiety, self-harm, and suicide aligns with smartphone-plus-algorithmic-feed saturation around 2012); the cross-national synchrony (the same inflection across the US, UK, Canada, Australia, the Nordics — hard to pin on any one country's economy or politics); and dose-response plus the gendered pattern (heaviest users, especially girls, are worst affected). He explicitly considers and dismisses the obvious rivals — the 2008 crash, academic pressure, the news cycle — on the grounds that none are synchronized across countries the way the tech rollout was.
The prescription. Four norms, which he claims could largely reverse the damage in about two years given collective action across tech, government, schools, and parents: no smartphones before high school; no social media before 16; phone-free schools; far more unsupervised play and independence. (Australia's under-16 ban is essentially norm #2 written into law.)
Where it's genuinely strong
The displacement logic (kids really did lose sleep, play, and in-person time, and that's developmentally real whatever is on the screen); the collective-action point (it really is brutal for one family to opt out alone); and the cross-national timing correlation (a striking pattern that rival explanations struggle to match).
Where it bears its weight — the seams your argument enters through
First, correlation to causation: the timing is striking, but most of the underlying research is correlational, and the experimental evidence is thin and largely run on adults 18–35, not the children the policy targets (Odgers, Orben, Przybylski). Second, device-as-cause vs. device-as-delivery-vehicle: his remedies treat access as the lever; your claim is that the harm lives in extractive design and intent, which predates and will outlive any given device — cut the access and the same design finds a new surface. Third, reversibility: the two-year claim only holds if the phone is the cause; if the real driver is a broader configuration (consumerization, design-from-mistrust), removing phones leaves the engine running. Fourth, capacity: even granting every harm, delay-and-restrict produces ignorance, not competence — the gun-safety point.
The concession worth making out loud
He's right that something cracked, right that it cracked around 2012, right that it isn't teenagers' imagination, and right that individual willpower can't beat a collective-action trap. Granting all of that costs you nothing — and it earns you the standing to dispute what did the cracking and what to do about it, which is the whole argument.