A Minimum Age Online

Bar under-13s from social media accounts and switch off algorithmic feeds for everyone under 17.

AI evaluation · not yet reviewed by a human

This evaluation was produced and sourced by an AI model; a human review is still pending. Figures and conclusions may still change. The review log is at the foot of the page.How review works →

Platforms already forbid accounts for under-13s in their terms of service and do almost nothing to enforce it. The bill before Congress would make the age limit a legal duty: a platform that knows, or from the circumstances plainly should know, that a user is under 13 must close the account, and it must switch off algorithmically personalised recommendations for every user it knows to be under 17, leaving them a feed of what they chose to follow. It requires no age verification and no new data — platforms act on what they already hold — and it covers only services that live mainly on advertising and host a community forum. A second title makes schools that receive federal broadband subsidies block social media on school networks. Australia went further in December 2025, barring under-16s from ten named platforms, and removed 4.7 million accounts in the first week. This evaluation compares five years under the American bill against five years without it.

Balance

Much better for the future · 0.91 previous scale

Balance on the previous scale. The Bilanz 2.0 simulation is not yet available for this evaluation. The category comes from the share of the debate on the pro side (r).

For 46 · 91 % Against 4.8 · 9 %

Size class: medium Scale of this evaluation: Normalised Impact — unitless, calibrated to this topic. For comparison: one point here is worth roughly 200 million euro per year. The two largest figures here rest on one measurement — what Facebook's arrival did to college students twenty years ago — carried across to today's feeds at 30 percent, and both are cut to the accounts a platform actually knows to be young, because the bill requires no age check. The measurement is a net one: what the feed gave those students is already inside it, which is why the loss of connection stands on this page without a figure of its own. How we score →

Arguments for

Arguments against

6 arguments evaluated · Scoring v1.3 Δ absolute +41.2

Arguments — For

2 arguments

Better years at school

27of 100

The same study that found worse mental health found more students reporting that it interfered with their academic work. That is a second consequence rather than the same one told twice: what a person learns between thirteen and seventeen shows up in their earnings for forty years.

Value 8 · Life chancesImpact 8.6Plausibility 4

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is what a young person learns and what that is worth to them later: qualifications gained, courses not dropped, the difference between finishing and not. This site places it in the class it uses for subsistence and life chances, above economic output and below life and health. It is genuinely separate from the mental health counted above — a student can be unhappy and still learn, or content and still fall behind — and it is counted only to the extent the evidence links the two. What the schooling is worth to the wider economy rather than to the person is not counted, because it would be the same hours twice. The value sits in the upper part of the scale, at the level this site uses for subsistence and life chances.

About 17 million Americans are of secondary school age, and the same 70 percent as in the previous argument are reached by the feed rule. The study behind that argument found that Facebook's arrival raised the share of students reporting that poor mental health had impaired their academic work from 13 to 16 percent ^[4]^[9]; 30 percent of that is carried across for a feed switched off rather than a platform removed, so about 0.9 percent of pupils, roughly 107,000, are newly spared such impairment each year, in a range from 0.3 to 1.8 percent. What an impaired pupil loses is a construction: a tenth of a school year's learning is assumed, in a range from a twentieth to a quarter. A year of American schooling is worth roughly eight percent of lifetime earnings, and lifetime earnings are taken at 2 million euro — forty working years at 50,000 — without discounting, as this site does everywhere; a tenth of a year is therefore worth about 16,000 euro to the pupil concerned, and the 10,700 pupil-years a year come to about 1.7 billion euro. The figure counts each year's pupils once and follows their earnings for a working life, which is how this site books schooling everywhere. The Impact is the largest in this debate, and it is large because a small share of a school year, followed through forty years of earnings, is worth a great deal — the one step that carries it, from self-reported impairment to learning lost, is the one nobody has measured.

▸ Show calculation ▾ Hide calculation

	Americans of secondary school age		17 million students
×	Share whose platform knows their age the same reach as in the first argument: no age check is required, platforms act on the age they hold [6] [6]	70 %	11.9 million students
×	Pupils newly spared mental-health impairment of their school work Setting, range 0.3 to 1.8 percent: Facebook's arrival raised the share reporting such impairment from 13 to 16 percent, and 30 percent of that is carried across for a feed switched off rather than a platform removed [4][9]	0.9 %	107,100 students
×	Learning lost by an impaired pupil Setting, range 5 to 25 percent: the source reports impairment, not grades	10 % of a school year	10,710 student-years
×	Lifetime earnings from a year of schooling a year of American schooling is worth roughly eight percent of lifetime earnings; forty working years at 50,000 euro, undiscounted as everywhere on this site	8 % × 2 million euro	1,714 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	8.57

Score 8.57 Impact × 8 Value × 4 Plausibility ÷ 10 = 27 of 100

The link is real in the source and thin in the application. The counterfactual is the same students at colleges without Facebook, and the academic finding rests on the same staggered rollout that carries the mental health result, so its identification is equally good ^[4]. What is much weaker is everything after it. The study measures students saying their work suffered, not grades or completion, and self-reported impairment and actual learning are not the same quantity. The step from there to lifetime earnings runs through a return-to-schooling figure estimated on whole years of education rather than on marginal quality within a year, which is a different object. The counter-mechanism is unaddressed: a student who spends less time on a feed does not necessarily spend it studying, and the American time-use evidence on what displaces what is not settled. Reverse causation does not arise in the underlying design. Two things are new since the first version: the share of pupils spared impairment now comes from the study's own three percentage points rather than a round one percent, and the loss per impaired pupil is a stated tenth of a year. The Plausibility is below the middle: the finding it starts from is identified and every step after it is assumed.

evidence basis: Study · P ceiling 8 identification: Quasi-experimental · rung ceiling 8 band: Chain closed, unevidenced · P 4–5

Counterfactual: American colleges without Facebook, same staggered rollout as pro-1 [4]. Design: quasi-experimental for the academic-impairment finding; everything after it is construction. Confounder: displaced time not going to study, unaddressed. Direction: no reverse causation in the underlying design. Ceiling: quasi-experimental 8.0 binds; a context transfer of 4.0 applies because the source measures self-reported impairment among college students and this argument prices learning among school pupils, valued with a return-to-schooling figure estimated on whole years. Band: chain closed but unevidenced — the chain is named and the displacement counter-mechanism is stated; only the measurement is missing.

Nothing measured argues against the claim; what is missing is any measurement of learning rather than of self-reported impairment. The counter-mechanism — that time off a feed does not become study time — is named and unresolved. Read back: about half the time, removing the feed improves a school year by roughly the amount assumed here.

Open: Australia's ban applies to a whole cohort at one date. Comparing school outcomes for Australian 15-year-olds against a comparable country over three years would measure this directly and could carry P to 6.

Adolescents in better mental health

19of 100

The best causal evidence on this question comes from Facebook's arrival at American colleges one at a time between 2004 and 2006. Where it arrived, symptoms of depression and anxiety rose, and the mechanism the authors identify is unfavourable comparison with other people. The bill removes the part of the product that does the comparing.

Value 9 · HealthImpact 4.7Plausibility 4.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is the mental health of people between about eleven and seventeen: how much they sleep, whether they are anxious in a way that interferes with school, how they feel about their own life against everyone else's. This site places it in the class it uses for life and health, one step below the top because adolescent low mood is generally recoverable and because what is measured is a symptom index rather than a diagnosis. The weight does not rise because the people concerned are young; what youth changes is how long the consequences run, and that belongs to the Impact. The academic consequences are counted separately in the next argument. The value sits one step below the maximum: the stream is health, and health that can be regained.

About 17 million Americans are aged 13 to 16, and roughly 4 million under-13s hold accounts anyway. The bill requires no age check: a platform acts on the age it already holds or on what the circumstances make obvious ^[6]. Around a third of British children with a profile carry an adult date of birth ^[7], and platforms now run age estimation on accounts that look younger than they claim, so about 70 percent of 13-to-16-year-olds are taken as reached by the feed rule, in a range from 50 to 90, and 40 percent of under-13s as actually losing their accounts, in a range from 15 to 70 — an under-13 account exists only because of a false date of birth, and it goes only when the platform can no longer pretend not to know. The size of what a feed does comes from the one clean measurement available: Facebook arrived at American colleges at different dates, and where it arrived, an index of poor mental health rose by 0.085 standard deviations — about 22 percent of the effect of losing one's job ^[4]^[9]. Removing algorithmic recommendation is not the same as removing the platform, so 30 percent of that effect is used, in a range from 10 to 60 percent: about 0.025 standard deviations across the 11.9 million teenagers reached, plus a larger effect on the 1.6 million under-13s who lose accounts entirely. A standard deviation on such an index is worth roughly 0.06 quality-adjusted years, in a range from 0.03 to 0.12. Together that is about 24,000 quality-adjusted years a year. The Impact is the largest health figure in this debate, and it is large because of how many adolescents there are rather than how much happens to each of them — and it is about half what it would be if the law reached every account.

▸ Show calculation ▾ Hide calculation

	Americans aged 13 to 16		17 million people
×	Share whose platform knows their age Setting, range 50 to 90 percent: the bill requires no age check and acts on the age a platform already holds [6]; a third of children carry an adult date of birth [7], and platforms now estimate age on accounts that look younger [6][7]	70 %	11.9 million people
×	Improvement in symptoms of poor mental health Setting, range 10 to 60 percent of the measured effect: Facebook's arrival at a college raised the index by 0.085 standard deviations, and this measure removes one feature rather than the platform [4][9]	0.025 standard deviations	297,500 standard-deviation-people
+	Under-13s who actually lose their accounts Setting, range 15 to 70 percent removed: an under-13 account exists only under a false date of birth and goes only when the platform can no longer pretend not to know; a larger effect per head, because the account goes rather than the feed [6][7]	4 million × 40 % × 0.06 standard deviations	393,500 standard-deviation-people
×	Quality-adjusted years per standard deviation Setting, range 0.03 to 0.12	0.06	23,610 quality-adjusted years
×	Value of the years the value of a healthy life year used across this site	40,000 euro each	944 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	4.72

Score 4.72 Impact × 9 Value × 4.5 Plausibility ÷ 10 = 19 of 100

This is the most contested empirical question in the area and the evidence divides by design rather than by conclusion. The counterfactual in the study carrying this figure is colleges that had not yet received Facebook, compared against those that had, on the same survey instrument before and after ^[4]. The confounder that would otherwise dominate — that early colleges differ from late ones — is what the staggered rollout absorbs, and reverse causation cannot arise, since a student's mood did not determine when Facebook expanded. Against that, meta-analyses of the wider literature repeatedly find effects near zero, and they are right to: most of that literature is cross-sectional and measures who uses social media rather than what it does. Two things limit the transfer here. The study is about college students in 2004 and this measure concerns adolescents in 2026, whose product is a video recommendation engine rather than a list of friends' profiles. And the study measures a platform arriving, while this measure removes one feature of it — which is why only 30 percent of the effect is carried across. The Plausibility is below the middle: the design behind the number is sound, and almost everything about the transfer to today's product is assumed.

evidence basis: Study · P ceiling 8 identification: Quasi-experimental · rung ceiling 8 band: Chain closed, unevidenced · P 4–5

Counterfactual: American colleges that had not yet received Facebook, on the same survey instrument. Design: quasi-experimental — generalised difference-in-differences on a staggered rollout (Braghieri, Levy and Makarin, American Economic Review 2022 [4]). Confounder: early-adopting colleges differing from late ones, absorbed by the staggered timing. Direction: no reverse causation, student mood did not determine Facebook's expansion order. Ceiling: quasi-experimental 8.0 binds below the studie ceiling of 9.0; a context transfer of 3.5 applies for two separate gaps — college students in 2004 against adolescents in 2026, and a platform arriving against one feature being switched off. The size doubt sits in the 10 to 60 percent band, not in P.

The chain is named and its one measurement is good: a staggered rollout, a tested comparison, no reverse causation. What is missing is a measurement on today's product. The meta-analyses that find effects near zero are not a counter-finding on the same outcome, because they are cross-sectional and measure who uses social media rather than what it does — a different question under a design that cannot answer this one. Read back: about half the time, switching off algorithmic feeds for the teenagers a platform knows about improves their mental health by roughly the amount assumed here.

Open: Australia removed 4.7 million under-16 accounts on a single date in December 2025. A comparison of adolescent mental health measures against a similar country over three years would replace the whole transfer with a direct measurement and could carry P to 7.

Arguments — Against

4 arguments · top 3 shown

Platforms rebuild at their own expense

2.8of 100

Closing every account a platform knows belongs to a child, deleting its data within ninety days, and running a second, unpersonalised feed for every user it knows to be under seventeen is engineering, legal review and moderation staff. The large platforms can pay for it easily, which does not make it free: the hours and the machines are spent either way.

Value 5 · Compliance costImpact 1Plausibility 5.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money spent by companies on compliance: engineers' time, legal review, moderation staff, the servers that run a second feed. This site places corporate spending at the middle of the scale, at the weight it gives to money as such — neither raised because the platforms are large nor lowered because they can afford it. The first version of this evaluation left the cost out on the ground that the platforms can carry it and it changes nothing about what they offer; that confuses who pays with whether anything is spent. What the spending buys — the protection of adolescents — is counted on the other side and is not netted here. The value sits at the middle of the scale: the stream is corporate money, and money is money whoever holds it.

The bill takes effect a year after enactment and reaches every service that collects personal data, lives primarily on advertising and hosts a community forum ^[6]. For each of them it means three pieces of work: finding and closing the accounts it knows belong to under-13s and deleting their data within 90 days; building and running a recommendation system for known teenagers that uses nothing about them but device, language, location and age; and the legal and moderation apparatus to show a regulator that it did not know what it should have known. The largest platforms have already built parts of this — teen accounts, age estimation, a chronological feed option — so what is counted is the increment. Ten large platforms at 15 million euro a year each, in a range from 5 to 40, and a long tail of a thousand mid-sized services at 50,000 euro a year give about 200 million euro a year over the five years, in a range from 60 to 600 million. That is a price set rather than found: no platform publishes what its teen-safety engineering costs. The Impact is a fifth of the health gain, and it is the one cost in this debate that arrives whether or not the law achieves anything.

▸ Show calculation ▾ Hide calculation

	Large covered platforms [6]		10 platforms
×	Yearly increment for closing known child accounts, an unpersonalised teen feed and compliance Setting, range 5 to 40 million: teen accounts, age estimation and a chronological option already exist, so only the increment counts	15 million euro each	150 million euro
+	Long tail of mid-sized covered services Setting: legal review and account closures without a dedicated team	1,000 × 50,000 euro	200 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	1

Score 1 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 2.8 of 100

That a legal duty of this shape costs engineering and legal time is not a prediction; what is uncertain is the size, and that sits in the range rather than here. The counterfactual is the current position, in which the same platforms already spend on teen accounts and age estimation voluntarily, so part of the work is done and would be done without the law — which is why the increment rather than the whole is counted, and why the range runs low. The one thing that could make the figure vanish is a platform simply switching every user under seventeen to a chronological feed it already offers, at almost no cost; the bill's exceptions are written to allow that, and a platform whose product is the personalised feed will not do it. No behavioural step carries the quantity, and reverse causation does not arise. The Plausibility is just above the middle: the spending is certain, the increment over what platforms already do is assumed.

evidence basis: Projection · P ceiling 6 identification: Definitional · no rung ceiling

Counterfactual: the current position, in which platforms already spend on teen accounts and age estimation voluntarily. Design: definitional — a duty to close known child accounts, delete their data and run an unpersonalised feed for known teenagers costs engineering and legal time [6]; no behavioural link carries the quantity. Confounder: spending that would occur without the law; handled by counting the increment and by the low end of the range. Direction: not applicable. Ceiling: projektion 6.0 binds; P sits just below it because the increment over voluntary spending is set rather than measured. The size doubt sits in the range 60 to 600 million euro.

Some people must prove who they are

1.7of 100

The bill requires no age check and no new data: a platform acts on what it already knows. But a platform that must not know is a platform that will want to be sure, and the ones that already run age estimation on accounts that look young will ask more people for a document. For those people, an account that never required an identity now has one attached.

Value 9 · Basic rightsImpact 0.4Plausibility 4.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is two things that arrive together for the people asked: the time spent proving their age, and the end of anonymous participation for anyone whose account is now tied to a document. This site places the second in the class it uses for the constitutional core and treats the first as time under compulsion, which sits near the top of the scale as well; the combination is priced just below life and health. The point is not that identity checks are wrong — banks do them daily — but that a social platform is where a great deal of political speech now happens, and an account connected to a document changes what can be said on it. Nothing here is counted for the risk of that record leaking, which is real and unquantified. The value is high because the stream is anonymity in public speech alongside hours of compelled time, not a convenience — and it is priced per person asked, however many that turns out to be.

Roughly 250 million Americans hold at least one social media account. The bill does not make anyone check them: it says in terms that nothing in it requires age verification or the collection of any age data a platform does not already hold, and it lets a platform be found to know a user's age only from what the circumstances make obvious ^[6]. What remains is the platforms' own response to that standard: a platform that is liable for what it should have known will want to be sure, and Meta already routes accounts its models flag as younger than declared into a verification step ^[10]. Ten percent of account holders are taken as asked to prove their age at some point in the five years, in a range from 2 to 40 — flagged adults, people whose accounts change hands or devices, users of platforms that decide to check everyone rather than argue about what was obvious. Each check is put at 12 minutes, in a range from 4 to 30, valued at the rate this site uses for time a person must spend with nothing in return: about 34 million euro a year. The larger part is the loss of anonymity for those checked, which has no market price: 2 euro a person a year is used, in a range from 0.5 to 10, the order of magnitude that surveys of willingness to pay for privacy return. Together that is about 84 million euro a year. The Impact is a tenth of what it would be if everyone had to be checked, which is what the first version of this evaluation assumed and the bill does not say.

▸ Show calculation ▾ Hide calculation

	Americans holding at least one social media account		250 million people
×	Share asked to prove their age within five years Setting, range 2 to 40 percent: the bill requires no check and no new data [6]; what is counted is platforms hedging a knowledge-fairly-implied standard — flagged adults, changed devices, platforms that decide to check everyone [10] [6][10]	10 %	25 million people
×	Time verifying and re-verifying Setting, range 4 to 30 minutes: uploading a document, waiting, retrying when a face scan fails	12 minutes a year	5 million hours
×	Value of forced time the rate this site uses for time a person must spend with nothing in return	6.85 euro an hour	34.25 million euro
+	Value set on the end of anonymous participation for those checked Setting, range 0.5 to 10 euro a person: the order of magnitude surveys of willingness to pay for privacy return; set here rather than found	25 million × 2 euro	84.25 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	0.42

Score 0.42 Impact × 9 Value × 4.5 Plausibility ÷ 10 = 1.7 of 100

That some people will be asked to prove their age is a prediction about how platforms answer a liability standard, not arithmetic. The counterfactual is the current position, in which platforms act on a self-declared date of birth and, increasingly, on their own age estimation. The bill's text is the strongest evidence against a large figure: it excludes any duty to verify and any duty to collect new data ^[6]. What carries the figure is a mechanism — the standard of knowledge fairly implied from the circumstances makes it costly not to know, and the commentary on the bill expects platforms to build verification steps for that reason ^[10] — plus the observed fact that the largest platform already sends flagged accounts to a check. The confounder is that this would happen anyway: Meta's age estimation predates the bill, and part of the checking counted here would occur under no law at all, which would cut the figure. The privacy half of the quantity is a stated price rather than a measurement. Reverse causation does not arise. The Plausibility is below the middle: the mechanism is named and partly visible already, and how far platforms go beyond what the bill demands is assumed.

evidence basis: Projection · P ceiling 6 identification: Mechanistic · rung ceiling 6 band: Chain closed, unevidenced · P 4–5

Counterfactual: the current position — self-declared dates of birth plus the platforms' own age estimation. Design: mechanistic — the bill requires no check [6]; the quantity rests on platforms hedging a knowledge-fairly-implied standard with verification steps [10], a behavioural link, not a legal one. Confounder: age estimation and verification prompts that platforms would run without the bill; unresolved. Direction: not applicable. Ceiling: projektion 6.0 and mechanistic 6.0 bind; P sits below because the share of people asked and the price on anonymity are both set rather than found. Band: chain closed but unevidenced — the mechanism is named and partly observed, the share is not.

Nothing measured argues against the claim: platforms do already ask flagged users for documents. What is missing is any measure of how many more would be asked under the bill's knowledge standard than are asked today. Read back: about half the time, the bill leads platforms to check roughly one account holder in ten over five years.

Open: Once the law is in force, the platforms' transparency reports would show how many accounts were sent to a verification step and how many adults were misclassified; that would replace the assumed share with a count and could carry P to 6.

Small sites close rather than comply

0.3of 100

When the United Kingdom imposed online safety duties in 2025, a number of small British forums shut down rather than carry the legal and compliance load. The American bill asks far less of them — no age checks, no risk assessments, and no duty at all unless a site lives mainly on advertising — but a volunteer-run board that is covered still faces a federal liability it cannot price.

Value 7 · Public debateImpact 0.1Plausibility 3.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is places for people to talk to each other that stop existing: a forum for a rare disease, a local history board, a community that never had a business model. This site places that in the class it uses for participation and the conditions of public debate, above money and below health. What is priced is the loss of the venue rather than any particular conversation, and no attempt is made to weigh a small forum against a large platform by size — the argument is about the ones for whom no substitute exists. The compliance cost to large platforms is not counted here, because they can carry it and it changes nothing about what they offer. The value sits in the upper middle of the scale, at the level this site uses for the conditions of public debate.

The British experience since 2025 is the closest precedent: hobby forums such as the London fixed-gear cycling board closed rather than meet the Online Safety Act's requirements, and support communities put government-ID checks in front of their doors ^[8]. Very little of that regime is in this bill. It requires no age assurance and no new data ^[6]; it covers only a service that collects personal data, lives primarily on advertising or data sales, and hosts a community forum — so a board with no advertising is outside it, as are wikis, review sites, gaming, email and direct messaging; and what a covered board must do is close accounts it knows belong to under-13s and not run a personalised recommendation system for known teenagers, which a plain forum does not run. What remains is liability under a standard of knowledge fairly implied from the circumstances, enforced by the Federal Trade Commission and state attorneys general, for an operator with no lawyer. Twenty million euro a year is used for the value of what closes on that account, in a range from nothing to 500 million; the first version of this evaluation carried 150 million, scaled from the British closures as if the duties were alike. The Impact is the smallest in this debate, and it is small because the bill's definition of a covered platform and its refusal to mandate checks leave most small venues untouched.

▸ Show calculation ▾ Hide calculation

	Value of small forums and community sites that close rather than carry the liability Setting, range 0 to 500 million euro: British boards closed under a regime with risk assessments, complaint systems and age assurance [8]; this bill imposes none of those, requires no checks, and covers only ad-funded services with a community forum [6] — what remains is an unpriceable federal liability for volunteer operators [6][8]		20 million euro
÷	Normalised Impact scale of this evaluation	200 million euro a point	0.1

Score 0.1 Impact × 7 Value × 3.5 Plausibility ÷ 10 = 0.3 of 100

The mechanism has a documented precedent under a different law and no measurement under this one. British forums did close after online safety duties took effect, which establishes that small venues do shut rather than carry compliance they cannot price ^[8]. The counterfactual there is the same forums before the duty, which is weak — several were already struggling and the duty was the occasion rather than the cause for some of them, and nobody has separated the two. The transfer is the larger problem: the British regime imposed risk assessments, complaint systems and age assurance; this bill imposes none of those and covers only ad-funded services with a community forum ^[6], so the chain that closed British boards mostly does not exist here. What is left is fear of a federal liability, and the precedent for that — the children's privacy law of 1998, which also turns on knowledge — did not close forums; it produced a line in their terms of service. Reverse causation is a live concern for the British closures. The Plausibility is well below the middle: the effect is documented under a heavier law, and under this one the closest precedent says it mostly does not happen.

evidence basis: Precedent · P ceiling 6 identification: Mechanistic · rung ceiling 6 band: Load-bearing link disputed · P 3.5–4

Counterfactual: the same British forums before the Online Safety Act duties — weak, since several were already struggling [8]. Design: mechanistic — a documented precedent under a different, heavier regime, without a design separating the duty from prior decline. Confounder: forums closing for unrelated reasons and attributing it to the duty; unresolved, and reverse causation is live. Direction: contested for the British closures. Ceiling: mechanistic 6.0 binds below the praezedenz ceiling of 8.5; P sits well below it because the American bill lacks the duties that closed British boards and the closest American precedent, a knowledge-based children's privacy law, did not close forums. Band: load-bearing link with a finding against it — the closest American precedent, a knowledge-based children's privacy law, did not close forums.

British forums did close after online safety duties took effect, so the direction is documented under a heavier law. The link that carries the quantity here — that a knowledge standard without age checks closes American forums — has a finding against it: the 1998 children's privacy law, built on the same kind of standard, produced a line in forums' terms of service and no wave of closures. Read back: about a third of the time, a duty of this kind closes small venues worth roughly the amount assumed here.

Open: The bill's definition of a covered platform is now in the text [6]; what would settle the rest is a count of American forums that close or geoblock within a year of the law taking effect, against the British count under the Online Safety Act.

Teenagers lose what they get from it

0of 100

The same feed that produces unfavourable comparison also produces the group chat, the club, the person who has the same rare illness, and for teenagers who are isolated where they live, the only people who are like them. Removing it removes both. Nobody has measured the second half — but the measurement the first argument rests on is a net one, and the loss is already inside it.

Value 9 · HealthImpact 0Plausibility 4

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is the same one the first argument counts, running the other way and on the same people: connection, belonging and support, and their absence. It would carry the same weight for the same reason — what is at stake is adolescent mental health, one step below the top of the scale because it is recoverable. Nothing here is counted for what parents gain or lose, which is a different stream on different people and is not priced anywhere in this evaluation. The argument is kept on the page because readers reach for it first, and because the reason it carries no figure of its own is the most important thing to understand about this debate. The value would sit one step below the maximum; it is not applied, because the stream is already netted inside the first argument.

The same adolescents are affected, and the effect runs the other way. Its size cannot be booked here, because the figure in the first argument is not the harm of a feed but the net of harm and benefit: the study measured how students' mental health changed when Facebook arrived, with everything the platform gave them — the friends kept, the groups found — already inside the number ^[4]^[9]. The net was negative, which is what says the harm outweighed the benefit; subtracting the benefit again on this side would count it twice, and the first version of this evaluation did exactly that at two fifths of the gain. What is not inside the measurement is the distribution: the loss falls hardest on adolescents who are isolated offline — in rural areas, with rare conditions, with an identity unwelcome where they live — for whom the feed is not one social channel among many. Whether the average net effect measured on college students holds for them is an open question, not a figure. The Impact is zero by construction: what this argument describes has already been taken off the other side's number, and the one thing it might add — a different balance for isolated teenagers — has never been measured.

This argument is grounded in the same study as the first one and reads it from the other side. The counterfactual is identical: colleges before and after Facebook arrived ^[4]. What that study measures is a net effect, so the benefit side is inside it rather than separate from it, and no design isolates the two components — which is why nothing can be added here without counting the benefit twice. The confounder in the qualitative literature that does examine the benefits is severe: teenagers who say social media helps them are describing their own experience, and the ones most likely to say so are the ones most attached to the product. Reverse causation is unresolvable there. What can be said is that the net effect being negative bounds the benefit from above — it cannot exceed the harm, or the measured sign would flip. The Plausibility is below the middle and is not applied: the stream certainly exists, and it has already been netted where the measurement was made.

Excluded (axis = 0): Already inside pro-1: the college measurement that carries the health gain is a net effect — what the platform cost students after what it gave them — so the connection lost is subtracted there once and cannot be subtracted again here. Set to zero on 14.09.2026; the first version booked it at two fifths of the gain on top.

evidence basis: Mechanism · P ceiling 6 identification: Mechanistic · rung ceiling 6 band: Chain closed, unevidenced · P 4–5

Counterfactual: the same colleges before and after Facebook arrived — the same source as pro-1 [4], which measures a net effect and therefore cannot separate benefit from harm. Design: mechanistic — the benefit component is inferred, not identified. Confounder: self-reported benefit correlating with attachment to the product; unresolvable in the qualitative literature. Direction: reverse causation is live for the self-report evidence. Ceiling: mechanistic 6.0 binds. Excluded from the total: the stream is inside pro-1's net measurement (zero_reason).

Nothing measured argues against the claim; what is missing is any design separating the benefit of a feed from its harm. The net effect being negative bounds the benefit from above, and the net is what the first argument already carries. Read back: the connection teenagers lose is real and is already subtracted on the other side of this page.

Open: Australia removed 4.7 million under-16 accounts on one date. A comparison of loneliness and support measures for Australian 14-year-olds against a comparable country would give the first direct reading of this side of the ledger.

Summary

Almost everything in this debate is one number seen from two sides: what an algorithmic feed does to the adolescents a platform knows it has. The best evidence for the harm is genuinely good — Facebook arrived at American colleges one at a time, and where it arrived, mental health measurably worsened — but it is twenty years old, it concerns a different product, and it measures a platform appearing rather than a feature being switched off. It is also a net measurement: whatever the feed gave those students is already inside the figure, so the connection teenagers would lose cannot be subtracted a second time, and the first version of this evaluation did. The bill itself is narrower than its critics' picture of it: it mandates no age verification and no new data, it acts on what a platform already knows or should obviously know, and it excludes services that do not live on advertising — which removes most of the identity-check cost and most of the small-forum closures from the ledger and leaves the platforms' own rebuilding costs and whatever checks they choose to run to stay on the right side of a knowledge standard. What remains open is how many accounts the law actually reaches, and whether a self-reported impairment of school work in 2005 turns into learning lost in 2027.

Outlook — effect over time

Much better for the future · 0.91 previous scale

Sources

Route Fifty: How would proposed age restrictions on social media use actually work?. route-fifty.com
Senator Brian Schatz: Kids Off Social Media Act. schatz.senate.gov
eSafety Commissioner, Australia: Social media age restrictions. esafety.gov.au
Braghieri, Levy and Makarin, American Economic Review 112(11): Social Media and Mental Health. aeaweb.org
Pew Research Center: Teens, Social Media and AI Chatbots 2025. pewresearch.org
U.S. Government Publishing Office: S. 278 — Kids Off Social Media Act (119th Congress), bill text. govinfo.gov
Ofcom: A third of children have false social media age of 18+. ofcom.org.uk
Techdirt: Didn't Take Long To Reveal The UK's Online Safety Act Is Exactly The Privacy-Crushing Failure Everyone Warned About. techdirt.com
Braghieri, Levy and Makarin, CEPR VoxEU: Social media and mental health. cepr.org
Truth on the Market: Can You Actually Keep Kids Off Social Media Without Age Verification?. truthonthemarket.com

Last reviewed by Claude Fable 5.1 · September 14, 2026 · 4× AI, not yet reviewed by a human

September 14, 2026AI reviewClaude Fable 5.1re-scored
v2 nach Gesamtprüfung 08.09.: Gesetzestext S. 278 verlangt keine Altersprüfung — con-1 0,84 → 0,084, con-3 0,15 → 0,02; con-2 (40 % des Netto-Effekts) als Doppelzählung auf 0; pro-1 mit Reichweite 1,6 → 0,94; pro-2 Lebenseinkommen korrekt 0,68 → 1,71; con-4 Plattformkosten neu. r 0,56 → 0,91.
September 6, 2026AI reviewClaude Opus 5record updated
i_spanne an allen 5 Argumenten aus den englischen Ketten, kein Transfer im Record. Kategorie kippt von Ausgeglichen (r 0,56) auf Schlechter (P(D>0) 0,15).
September 6, 2026AI reviewClaude Opus 5record updated
i_spanne an allen 5 Argumenten aus den englischen Ketten, kein Transfer im Record. Kategorie kippt von Ausgeglichen (r 0,56) auf Schlechter (P(D>0) 0,15).
September 6, 2026AI reviewClaude Opus 5First evaluation
Created for the English side: the harm is taken from the staggered Facebook college rollout, the benefit is booked as a share of it because nothing measures it.

Evaluations are produced with AI support and reviewed on a schedule for new developments; human passes are marked separately.How we review →