Yes, but…
Good objections deserve honest answers.
Common reasons people give for not taking this seriously — answered with evidence, not rhetoric.
01 I've got nothing to hide. +
Privacy protects far more than wrongdoing — it protects the right to form ideas without surveillance, have relationships without monitoring, and control your own narrative. The data assembled about you is also used to predict and influence what you will do next — not just record what you have done. (Solove, 2011)
02 I can just scroll past what I don't like. +
A fraction-of-a-second pause before scrolling still registers as engagement. The system records it. Your feed shifts. Ignoring a post does not take you out of the ranking that chose it — the harm is cumulative and environmental, not any individual post.
03 I know when I'm being manipulated. +
This is called the third-person effect (Davison, 1983) — a robust finding in media psychology, confirmed by a meta-analysis of 106 studies. People believe they are less susceptible than others. Education and intelligence do not protect against content calibrated for critical thinkers.
04 Social media is just entertainment. +
Entertainment is not neutral when it is where many people get their news, form political opinions and understand the world. The distinction between entertainment and information collapsed once social media became a main information channel for billions of people.
05 If I don't like it I can always just leave. +
The network effect makes this much harder than it sounds. Leaving WhatsApp or Instagram means losing access to your social network, not just an app. And the same streaks, notifications and variable rewards that keep you scrolling are working on you at the moment you consider leaving — the Gamification page shows how.
06 The algorithm shows me what I want to see. +
The algorithm shows you what keeps you watching — not what you want, but what you respond to. The distinction matters because outrage, fear, and tribal conflict keep people watching longer than content that makes them happy. Your preferences are also shaped by what you have been shown.
07 This is just how technology works. +
Aza Raskin, who designed infinite scroll, has publicly stated he regrets it. Tristan Harris, former design ethicist at Google, testified before the US Senate that the harms result from design choices, not technology itself. The people who built these features are not arguing that this is inevitable.
08 The media is just as bad — newspapers and TV chased attention long before the feed. +
True, and this site says so: the Journalism page sets out ownership concentration, advertiser capture, clickbait drift and outright fabrication. The incentive really is the same one — attention sold to advertisers. Four things are not. A newspaper prints one edition for everyone; a feed ranks a different one for each person, in real time, out of a profile built from what you did a moment ago. A front page reaches a country; Meta counted 3.60 billion people using its apps on an average day in June 2026. A publisher answers in law for what it prints, while online intermediaries were given a liability shield — Section 230 in the United States in 1996, the e-Commerce Directive in the EU in 2000 — so nobody carries an editor's responsibility for what the ranking chose. And a printed page or a broadcast is one object seen by everyone, so it can be checked afterwards; what a feed showed one person is visible to no one else, which is why the EU had to require the largest platforms to keep a public repository of the advertisements they present and to open their data to vetted researchers. What journalism has and the feed does not is a correction mechanism and a public record.
09 I'm educated. This doesn't affect me the way it affects others. +
Being good at reasoning is not the same as being protected from it. In a landmark experiment, the most numerate people were the most polarised when the same data touched their politics: greater skill was spent reaching the conclusion their side already favoured. Intelligence does not protect against content that confirms existing beliefs. (Kahan et al., 2017)
10 Children today are digital natives — they can handle it. +
Fluency is not the same as critical distance: digital natives are fluent in the interface, and fluency is not protection. An audit study found that accounts set up as boys and young men were recommended anti-feminist content they had not looked for. At the far end, Home Office figures show that 43 of the 232 terrorism arrests in Great Britain in the year to March 2025 — nearly one in five — were of children aged 17 or under; MI5 says terrorism now “breeds in squalid corners of the internet”. None of it looks like immunity.
11 Haidt has been debunked. +
Contested, not debunked — and the argument here does not rest on him. The critiques are real, and this site carries them: Candice Odgers' review in Nature argues that the timing evidence is weaker than The Anxious Generation presents; a meta-analysis by Christopher Ferguson and colleagues, pooling 33 studies published between 2015 and 2019, found screen media playing little role in mental health outcomes; and the standing objection — that nearly all of the evidence is correlational — is Orben and Przybylski's, which the Mental Health page cites. What survives those critiques is what the site actually leans on. Sleep is the best-documented route: across 20 studies of 125,198 children, using a device at bedtime was associated with too little sleep, and so was a device merely present in the bedroom. Facebook's own researchers saw the same signal, in internal surveys that are no stronger than the published studies, and the product continued: evidence about conduct, not about effect size. And two studies got closer to an experiment: randomly deactivating Facebook for four weeks improved a subjective-wellbeing index by 0.09 standard deviations — about a quarter to two-fifths of what psychological therapies achieve — and Facebook's staggered arrival at US colleges raised an index of poor mental health by 0.085 standard deviations. Small, one platform, and not the whole question. The position on the page does not move either way: the correlation is established, the causation is contested, and a live dispute is what a science with a correction mechanism looks like from the inside.
12 This sounds like a conspiracy theory. +
No conspiracy is required. The harms are emergent from commercial incentives — multiple actors optimising for their own goals. The primary sources are the platforms' own internal documents, official Senate investigations, and peer-reviewed research. Only incentives.
13 This was written by an AI. +
Partly, and the note at the foot of every page says so — whatever review that page has had. AI tools help find and read sources, draft and edit text, translate pages and check each other's work. They do not decide. A person chooses what is published and is answerable for it; a page a person has read carries that date in its footer, and a translation goes live only after a native speaker has reviewed it, unless its edition is published under volunteer review, which every page of it says. The method is built so that you do not have to take that on trust: every factual claim links to its source, each one carrying a number, a date, a named study or a quotation is tracked with how it was checked in the claim ledger, which is a page of this site, and every change made to the site is listed, with its date, in the change log. Mistakes still get through; when we find one, or a reader tells us, we correct it and log it on the corrections page. EU law requires this disclosure for AI-generated text published to inform the public, and exempts text that a person has reviewed and takes editorial responsibility for — we disclose anyway.
14 What can I actually do about it? +
Understanding the mechanism is genuinely protective. The Bad News game at getbadnews.com teaches six manipulation techniques in about 15 minutes — research finds the resistance lasts for months with repeated testing and fades within two months without it. The OECD's count of countries with social media age restrictions in force or under consideration rose from one, to twelve, to twenty-five in sixteen months; three of them had a law actually in force. Structural change is accelerating.
15 This is just a few bad actors — most people and platforms are fine. +
The harm is not produced by bad people. It is produced by an incentive structure. Platforms optimising for engagement, advertisers purchasing behavioural profiles, engineers building the most compelling features — none are individually malicious. The system produces harm emergently. That is what makes it harder to fix than bad actors would be.
16 Age checks just mean everyone has to hand over their ID. +
Not necessarily — though the worry is real, and this site does not treat it as settled. In the UK, Ofcom's guidance sets a standard it calls “highly effective age assurance” and names the methods that can meet it: open banking, photo-ID matching, facial age estimation, mobile-network operator checks, credit-card checks, digital identity services and email-based age estimation. A typed-in birthday does not qualify. Services pick among the methods, and each deployment is judged on technical accuracy, robustness, reliability and fairness. The EU is building something narrower: an age-verification blueprint published in July 2025, on the same technical specifications as the coming European Digital Identity Wallets, in which “online services will only receive a proof that the user is over 18, without any other personal details” — and each proof is used once, so it cannot be used to follow someone from one service to the next. Denmark, France, Greece, Italy and Spain are piloting it. What none of that settles is the objection underneath: a check strong enough to stop a determined 14-year-old generally has to run on every adult too, facial estimation means processing a face, and the infrastructure is new. The Age Restrictions page calls that problem genuinely unsolved and the evidence that the laws help still thin — nothing here resolves it.
17 Isn't moderating any of this just censorship? +
It is the objection that matters most, so it has its own page. The short version: “free speech” is asked to cover two different things — a person speaking, and a system deciding which of millions of messages billions of people see. The principle was built to protect the first. Whoever holds the power to moderate can abuse it, which is a real worry — so the honest question is not whether to sort content (a feed already does) but who decides, by what rules, and whether the decision can be seen and challenged. Two democratic traditions, the American and the European, answer differently — which tells you the line is a choice, not a law of nature.
The point of this page is not to win arguments. It is to make sure the objections have been genuinely heard and honestly answered.
Sources
- Solove, D. J. (2011), Nothing to Hide: The False Tradeoff Between Privacy and Security, Yale University Press (the publisher's page is for the 2013 paperback).
- Davison, W. P. (1983), The Third-Person Effect in Communication, Public Opinion Quarterly 47(1), 1–15.
- Sun, Y., Pan, Z. & Shen, L. (2008), Understanding the Third-Person Perception: Evidence From a Meta-Analysis, Journal of Communication 58(2), 280–300 — 60 papers, 106 studies; the perception is robust across research procedures.
- Kahan, D. M., Peters, E., Dawson, E. C. & Slovic, P. (2017), Motivated numeracy and enlightened self-government, Behavioural Public Policy 1(1), 54–86.
- MI5 — Director General Sir Ken McCallum, threat update (16 October 2025) — “one in five of the 232 terrorism arrests last year were of children under 17”; “Terrorism breeds in squalid corners of the internet”. The count is the Home Office’s, for the year to March 2025, where the same figure appears as 43 arrests (19%) of people aged 17 and under.
- Institute for Strategic Dialogue (2022), Algorithms as a Weapon Against Women: How YouTube Lures Boys and Young Men into the "Manosphere".
- UK Home Office (12 June 2025), Operation of police powers under the Terrorism Act 2000 and subsequent legislation: arrests, outcomes, and stop and search, Great Britain, quarterly update to March 2025 — 232 arrests in the year ending 31 March 2025, 43 of them (19%) of people aged 17 and under.
- OECD, Social media age restrictions for children: why they are rising and what comes next (April 2026).
- Andersson, H. (3 July 2018), Social media apps are ‘deliberately’ addictive to users, BBC News — Aza Raskin, who designed infinite scroll in 2006, “said he had not set out to addict people and now felt guilty about it”.
- US Senate Committee on Commerce, Science and Transportation (25 June 2019), Optimizing for Engagement: Understanding the Use of Persuasive Technology on Internet Platforms — Tristan Harris's testimony.
- Meta Platforms (29 July 2026), Meta Reports Second Quarter 2026 Results — the company's own count: “DAP was 3.60 billion on average for June 2026” (Family daily active people), as filed with the US Securities and Exchange Commission.
- Communications Decency Act, Section 230 (1996) — Legal Information Institute, Cornell; the US liability shield for what users post.
- European Union, Directive 2000/31/EC on electronic commerce — Articles 12–15: the hosting exemption and the absence of a general obligation to monitor.
- European Union, Regulation (EU) 2022/2065 (Digital Services Act) — Article 39, the public repository of advertisements presented by very large platforms, kept until a year after an advertisement was last shown; Article 40, data access for vetted researchers.
- Odgers, C.L. (2024), The great rewiring: is social media really behind an epidemic of teenage mental illness?, Nature 628, 29–30 — the main critical review of Haidt's book.
- Haidt, J. (2024), The Anxious Generation, Penguin Press.
- Ferguson, C.J., Kaye, L.K., Branley-Bell, D. et al. (2022), Like this meta-analysis: Screen media and mental health, Professional Psychology: Research and Practice 53(2), 205–214 — 37 effect sizes from 33 studies published 2015–2019.
- Orben, A. & Przybylski, A.K. (2019), The association between adolescent well-being and digital technology use, Nature Human Behaviour 3 — the correlational-evidence critique.
- Carter, B., Rees, P., Hale, L., Bhattacharjee, D. & Paradkar, M. S. (2016), Association Between Portable Screen-Based Media Device Access or Use and Sleep Outcomes: A Systematic Review and Meta-analysis, JAMA Pediatrics 170(12) — 20 studies, 125,198 children; bedtime use OR 2.17 for inadequate sleep, access without use OR 1.79.
- Allcott, H., Braghieri, L., Eichmeyer, S. & Gentzkow, M. (2020), The Welfare Effects of Social Media, American Economic Review 110(3), 629–676 — randomised four-week Facebook deactivation; the subjective well-being index improved by 0.09 standard deviations, “about 25–40 percent as much as standard psychological interventions”.
- Braghieri, L., Levy, R. & Makarin, A. (2022), Social Media and Mental Health, American Economic Review 112(11), 3660–3693 — Facebook's staggered roll-out across US colleges; the index of poor mental health rose by 0.085 standard deviations. Quasi-experimental, not randomised.
- The Facebook Files (Wall Street Journal, 2021) — Frances Haugen's internal disclosures, including the internal finding on teenage girls and body image.
- Ofcom (January 2025), Age checks to protect children online — the methods Ofcom considers capable of being highly effective, the exclusion of self-declared age, and the four criteria (technically accurate, robust, reliable, fair).
- European Commission (14 July 2025), Commission makes available an age-verification blueprint — built on the specifications of the European Digital Identity Wallets; “online services will only receive a proof that the user is over 18, without any other personal details”; each proof used once; piloted in Denmark, France, Greece, Italy and Spain.
- European Union, Regulation (EU) 2024/1689 (AI Act), Article 50 — Article 50(4) requires deployers to disclose AI-generated text published to inform the public on matters of public interest, and exempts text that has undergone human review or editorial control where a person holds editorial responsibility.
- Roozenbeek, J. & van der Linden, S. (2019), Fake news game confers psychological resistance against online misinformation, Palgrave Communications 5(65) — the Bad News evaluation, 15,000 players; the six techniques the game teaches, in about 15 minutes of play.
- Maertens, R., Roozenbeek, J., Basol, M. & van der Linden, S. (2021), Long-term effectiveness of inoculation against misinformation: Three longitudinal experiments, Journal of Experimental Psychology: Applied 27(1) — how long the effect holds.
Did this page help? Tell us what was wrong →