Haidt, Jonathan

social psychologist at NYU, author of The Coddling of the American Mind and The Righteous Mind

Jonathan Haidt: The Case Against Social Media

Lex Fridman Podcast · Jun 4, 2022

• Jonathan Haidt
• teen mental health decline
• social media and depression correlation
• dose-response versus childhood rewiring
• Tower of Babel metaphor
• David Shor cancellation
• Elon Musk Twitter acquisition
• content moderation versus platform architecture
• user authentication proposals
• polarization and democracy decline
• anti-fragility and free play

Core ideas

Haidt frames his intellectual trajectory as moving from studying left-right moral psychology (via The Righteous Mind) to studying what he calls a sudden, measurable discontinuity in both teen mental health and democratic institutions beginning around 2012-2014. He argues teen depression, anxiety, self-harm hospitalizations, and suicide (the latter two being behavioral, not self-reported, data) all show a sharp, sustained rise starting around 2010-2013 — not a gradual trend — hitting girls hardest and appearing similarly in the US, UK, Canada, Australia, and New Zealand, but less clearly in continental Europe. He explicitly rebuts a Zuckerberg claim (from Zuckerberg's own prior interview with Fridman) that Instagram is "generally positive" for teen mental health, and challenges the widely-cited comparison of social media's harm to "eating potatoes": he argues that comparison came from a study measuring general digital screen time (including Netflix and homework), not social media specifically, and that when researchers isolate social media use — and especially isolate girls — the correlation with poor mental health rises substantially (Haidt cites figures around r=0.15–0.22), which he argues is comparable to or larger than other accepted public-health risk correlations (e.g., childhood lead exposure and adult IQ, roughly r=0.09).

He argues the effect isn't best modeled as a "dose-response" (sugar-like) relationship but as a wholesale "rewiring of childhood": citing a decline in unsupervised free play beginning in the 1990s (predating social media, driven by parental fear), followed by Gen Z becoming the first cohort to get smartphones and social media accounts before puberty. Citing newer research (Amy Orben and Andrew Przybylski) suggesting harm is greatest for girls aged 11-13, Haidt argues there is no way to make platforms like Instagram or Twitter "safe" for pre-teens, and recommends raising the effective minimum age for such platforms to 16, with enforcement, rather than the current unenforced age-13 standard set by 1990s-era COPPA law. He distinguishes this business model specifically — ad-driven platforms where "content" from ordinary users drives engagement and users are "the product" — from general internet/smartphone use, arguing platforms like YouTube and TikTok are harder to assess (he states he lacks strong data on TikTok specifically) and that private messaging or multiplayer gaming don't show the same harm signature.

On democracy, Haidt argues social media (specifically Facebook and Twitter, in his framing) has pushed public discourse into a chronic "defend mode" (citing a simplified neuroscience model of approach/avoidance brain circuitry), undermining the "discover mode" cooperative problem-solving he associates with functioning pluralistic democracies. He uses the biblical Tower of Babel story as his organizing metaphor for his Atlantic essay ("Why the Past 10 Years of American Life Have Been Uniquely Stupid"): not a story of two sides at war, but of sudden, near-total fragmentation of shared language/narrative. He argues the distinctive, new element (beyond pre-existing polarization trends) is that a small number of highly effective users can use social media dynamics to intimidate and "cancel" people over specific words or framings rather than substantive disagreement — citing the case of political scientist David Shor, fired after tweeting a peer-reviewed study on the historical effectiveness of nonviolent versus violent protest. He frames this as a social-psychology "situational forces" problem rather than primarily individual moral failure, arguing that when the overwhelming majority of leaders in a given situation cave to pressure, the right diagnostic question is what about the situation makes that outcome so common.

Discussing Elon Musk's 2022 acquisition of Twitter, Haidt states an overall positive view of Musk's broader engineering achievements, but directly criticizes a specific tweet in which Musk announced he would vote Republican and accused Democrats of having "become the party of division and hate," arguing that as CEO of major companies, publicly declaring partisan allegiance in an inflammatory way violates a kind of professional/fiduciary norm for corporate leadership and further politicizes commercial life at a time Haidt says the culture war risks provoking serious civil unrest ("civil war," in his phrasing, which he says worries him more than a world war). Fridman pushes back, defending the tweet as an authentic, humor-inflected expression rather than calculated political messaging; Haidt maintains the concern is specifically about a CEO's fiduciary and normative responsibilities, not a broader claim that public figures must self-censor. Haidt separately credits a different Musk tweet — about authenticating platform users to filter out bots — as pointing toward what he considers the most important actual reform: shifting focus away from content moderation (which he argues has limited real impact, citing Frances Haugen's testimony that even aggressive moderation catches only a small fraction of harmful content) toward structural/architectural changes like verified human identity requirements and reputation-based visibility systems that make being "an asshole" carry a real social cost, analogous to physical-world consequences.

Haidt directly disputes Zuckerberg's claim (from the earlier Fridman interview) that "most academic studies... show that social media use is correlated with lower polarization," arguing this rests on selectively citing one study (Boxell, Gentzkow, and Shapiro) whose long-run (1970s-onward) trend lines obscure a sharper post-2012 polarization uptick visible in many (not all) countries' data, and noting independent measures of global democratic quality show a decline beginning in the late 2000s to mid-2010s. He references a broader collaborative research document he maintains with Chris Bale, organized around specific yes/no questions (e.g., does social media amplify emotional/false content, enable foreign interference, weaken trust, strengthen populism), as his preferred format for tracking contested evidence transparently. On potential solutions, he favors regulatory approaches (the U.S. Platform Accountability and Transparency Act, the UK's Age-Appropriate Design Code, a proposed Digital Platforms Commission) given what he calls a structural "social dilemma" preventing platforms from unilaterally raising age limits or reducing engagement-maximizing design without losing users to competitors, while acknowledging regulation risks being clumsy and prefers voluntary leadership where possible.

He closes with career and life advice built around three goals he teaches in an NYU course: becoming stronger (via voluntary exposure to real-world risk, fear, and setbacks, framed through an "anti-fragility" lens influenced by his free-play advocacy), smarter (via genuine exposure to disagreement and viewpoint diversity, citing his co-founding of Heterodox Academy), and more sociable (via perspective-taking), arguing that phones and overprotective parenting have jointly deprived young people of the developmentally necessary experiences of unsupervised conflict and struggle.

Excerpts