Reading Notes

Jonathan Haidt on Adjusting to Smartphones and Social Media

Episode: Jonathan Haidt on Adjusting to Smartphones and Social Media

Notes — Jonathan Haidt on Adjusting to Smartphones and Social Media

Notes on Jonathan Haidt in conversation with Tyler Cowen — Conversations with Tyler (https://conversationswithtyler.com/episodes/jonathan-haidt-anxious-generation/), 3 April 2024.

Four questions [Adler frame]

Q1 — What is it about as a whole? Haidt’s second CWT appearance, timed to The Anxious Generation, is a sustained adversarial argument with Cowen over whether smartphones and social media have caused a real, distinctive harm to adolescent mental health, or whether the pattern is better read as one more historical mood-shift that people will simply adjust to. After warm-up questions on parenting, moral foundations and COVID, and identitarianism, the conversation’s spine is Cowen’s repeated pressure-testing of Haidt’s causal story — the 2012 discontinuity, the cross-country evidence, the size of the correlations, the missing benefits, the COVID school-closure ‘natural experiment’ — before turning to Haidt’s proposed remedies: four norms (no smartphones before high school, no social media before 16, phone-free schools, more independent play) and the age-gating and liability policy needed to make them stick.

Q2 — How is it argued? Haidt argues as a social psychologist working systemically: he reads the 2012 discontinuity in mental-health and loneliness data as a collective-action problem, in which individual teenagers are trapped on platforms they would rather not use because leaving means being cut off from peers, and proposes coordinated norm-and-policy changes as the exit. His evidence is a synthesis of time-use surveys, PISA loneliness items, and correlational and longitudinal studies on social-media use and depression, curated with Jean Twenge and Zach Rausch. Cowen’s method is the same adversarial, concrete style seen with Coleman Hughes: he supplies counter-cases (chess improvement, Emergent Ventures teenagers doing ‘phenomenally well’, the Aztecs adapting to a diet without vitamin C) and asks Haidt to defend the causal story against each, rather than contesting the framework abstractly.

Q3 — Is it true, in whole or part? Haidt concedes ground on specific points — that the COVID school-closure period is ‘not a clean experiment’ because too many other things changed at once, and that the correlation between overall screen time and poor mental health is small (around 0.04–0.1) — while holding that the correlation rises to about 0.17 once the analysis is restricted to girls and social media specifically, a figure he and skeptics such as Andrew Przybylski and Amy Orben apparently now agree on. [?] That convergence is asserted in conversation rather than shown; the notes cannot verify it against the underlying studies. [?source] Cowen’s counter-evidence — the Anglosphere/Nordic concentration of the effect, Sabine Hossenfelder’s video citing very small variance-explained figures — is not fully rebutted; Haidt’s response leans on cross-country religiosity and secularism patterns (Eastern Europe improving, Protestant Europe worsening) that are asserted rather than walked through in any depth here. The exchange ends in partial, explicitly negotiated agreement (both concede the COVID natural-experiment point; both narrow to agreeing that 11–13-year-old girls on Instagram likely have a negative cost-benefit ratio) rather than a resolved empirical dispute.

Q4 — What of it? The episode is a case study in what a genuine, unresolved empirical disagreement between two serious interlocutors sounds like — useful precisely because Haidt does not simply restate the book’s thesis but has to defend it turn by turn against a sympathetic-but-skeptical questioner, conceding specific points while holding the larger claim. It extends Jonathan Haidt on Morality, Politics, Disgust, and Intellectual Diversity on Campus‘s treatment of moral foundations theory into a live application (the left/right gap in parenting and ‘rootedness’ against smartphone harm) and introduces the wiki’s first treatment of The Anxious Generation and the phone-based-childhood thesis.

Glossary

Phone-based childhood — Haidt’s term for a childhood organised around a smartphone and its constant connectivity, which he argues has displaced the ‘play-based childhood’ that human children, like other mammal young, need in order to develop socially and psychologically. [§ On whether AI can solve the screen time problem]

Experience blocker — Haidt’s description of the smartphone’s core effect: rather than simply adding a new activity, it displaces time that would otherwise go to talking with others in person, sleeping, being outdoors, and reading. [§ On whether AI can solve the screen time problem]

Collective-action problem (as applied to social media) — Haidt’s framing for why individual teenagers stay on platforms they say they would rather leave: each user needs to be on the platform to avoid social exclusion, so no one can unilaterally quit even though most would prefer, collectively, that the platform carried less weight in their lives. Cited to a Leonardo Bursztyn study finding students would pay to be collectively liberated from TikTok/Instagram, having earlier said they would need to be paid to give it up individually. [§ On the ease of adjusting to smartphones and social media]

The four norms — Haidt’s proposed remedy: no smartphones before high school (flip phones permitted), no social media accounts before 16, phone-free schools, and far more unsupervised childhood independence and free play (the Let Grow organisation, co-founded with Lenore Skenazy and Peter Gray). [§ On the ease of adjusting to smartphones and social media]

Age-gating — Requiring platforms to verify a user’s age before allowing an account, as bars and casinos are required to check ID; Haidt wants the legal age for signing a data-sharing ‘contract’ with a platform raised from 13 (set under COPPA) to 16, enforced, with an optional parental-permission carve-out. [§ On age-gating social media]

Key claims by section

On moral foundations theory as applied to COVID [§ On moral foundations theory as applied to COVID]

  • Haidt maintains social conservatives score higher on the purity/sanctity moral foundation, but argues COVID-era mask and distancing behaviour was driven less by germ aversion than by a proxy fight over government control and distrust of a medical establishment coded as aligned with ‘the blue team’.
  • He denies identifying with ‘the values of the left’; describes his own trajectory as moving from a younger, reliable Democrat identity to a self-described centrist position after writing The Righteous Mind, and further right of the median campus view after the post-2015 ‘cultural revolution’ on campus.

On psychological traits as a source of identitarianism [§ On psychological traits as a source of identitarianism]

  • Haidt distinguishes having an identity (which he says he has no problem with) from identitarianism: a theoretical commitment to putting identity, especially racial identity, first analytically, and treating identity groups as not morally equal (oppressors bad, oppressed good).
  • On psychological self-labelling (ADHD, autism, ‘my anxiety’), Haidt is agnostic empirically but suspects, drawing on Stoic and Buddhist ideas about interpretation, that identifying strongly with a mental-health label is probably bad for young people because it changes how they process experience.

On whether AI can solve the screen time problem [§ On whether AI can solve the screen time problem]

  • Haidt’s central claim: the harm from screens is not the screen time itself but the opportunity cost — displaced play, in-person friendship, sleep, and time outdoors — so an AI ‘digest’ that reduces raw screen time would not restore a play-based childhood, only re-engineer the phone-based one.
  • He predicts AI-mediated friends and romantic partners will be market-driven toward being ‘incredibly great for you’ with none of the friction of real relationships, which he calls a likely disaster for social development, since people need to learn to deal with difficult others.
  • Cowen presses that AI could cut the 5–10 hours per day Gen Z spends ‘servicing their network connections’; Haidt counters with the Bursztyn collective-action finding — students would pay to be collectively freed from TikTok/Instagram even though they wouldn’t individually quit — arguing AI intensifies rather than escapes the trap.

On school closures and well-being [§ On school closures and well-being]

  • Cowen’s challenge: if COVID lockdowns removed six to eight hours a day of in-person school contact yet didn’t much worsen youth well-being in Haidt’s data, doesn’t that undercut the claim that lost in-person time drives the harm?
  • Haidt’s rebuttal: American Time Use Survey figures already showed minutes spent with friends collapsing steeply from 2018–2019, continuing at the same rate through the COVID year with no inflection; school time itself had stopped functioning as ‘time with friends’ once phones entered classrooms (citing Jean Twenge/PISA loneliness-at-school data showing a post-2012 rise); and American schools are separately ‘depressogenic’ (Peter Gray’s data on suicide rates rising when school resumes), muddying lockdowns as a clean natural experiment.

On the ease of adjusting to smartphones and social media [§ On the ease of adjusting to smartphones and social media]

  • Cowen’s central challenge: humans have adjusted to agriculture, fire, and urbanisation, so why assume smartphones and social media are different — why not ‘get on with the adjustment’? Haidt’s answer: mammalian childhood has fixed developmental requirements (play, face-to-face interaction, synchrony) that a phone-based substitute cannot supply, unlike, say, an Aztec dietary workaround for scurvy.
  • Haidt sets out the collective-action framing (his NYU students report wanting to quit TikTok/Instagram but staying on because ‘everyone else is on it’) and the Bursztyn willingness-to-pay finding as the empirical anchor for treating this as a trap rather than a preference.
  • Haidt’s proposed exit: the four norms (no smartphones before high school; no social media before 16; phone-free schools; more independent play, via Let Grow) — offered as the coordinated solution to the collective-action problem, restoring the conditions under which kids could freely enjoy technology’s genuine benefits.

On age-gating social media [§ On age-gating social media]

  • Haidt wants the COPPA-era age of 13 (below which platforms cannot take a child’s data without parental knowledge) raised to 16 and enforced, with an optional parental-permission carve-out for younger teens; he distinguishes this from a blanket ‘no social media for minors’ ban and says he opposes government-mandated ID checks (citing Louisiana’s driver’s-licence requirement for pornography sites as a method he rejects).
  • He supports removing platforms’ Section 230 liability shield specifically for design choices (e.g. infinite scroll, constant refresh) engineered to be addictive, while preserving the shield for third-party content — arguing government created the current ‘zero guardrails, zero liability’ environment and can therefore fix it.
  • On enforcement realism, Haidt argues a platform achieving even 90–95% age-verification accuracy should earn some liability protection, versus the current baseline of essentially no enforcement (most middle schoolers already hold accounts); he analogises to shoplifting and Napster-era piracy, arguing that a genuine cost of violation (not necessarily a perfect one) meaningfully changes behaviour.

On the benefits of social media [§ On the benefits of social media]

  • Cowen presses the missing-benefits critique using Emergent Ventures teenagers who collaborate and ‘do science’ via Twitter/social media; Haidt responds that cited benefits nearly always describe older teens (16–18), that the internet (blogs, podcasts, multiplayer games, texting) already supplies the connective benefit without social media specifically, and that he has ‘never heard anyone make a good case’ that 11–13-year-olds benefit from social media.
  • Both converge on a narrower shared claim: the cost-benefit ratio for 11–14-year-old girls on Instagram/TikTok is probably negative; they leave 14–16-year-olds as the harder, unresolved case, with Haidt favouring an account age of 16.

See also