User:Louis/Manufacturing support for age verification

On June 15, 2026 the UK government announced a ban on under-16s using social media, expected to take effect early in 2027. The headline number behind it, the one the Secretary of State cited in a May 28, 2026 interview, was that 89% of parents supported a legal minimum age of access. That number is real, & the question behind it was fair. The problem is the gap between the question and the policy. Parents were asked whether they want a legal minimum age at all; the policy sets that age at 16 and bans everyone younger. Those are different questions, & the only figure in the consultation that speaks to 16 was shown only to the parents who had already said yes to a minimum age. I want to walk through how that gap gets papered over, using the UK consultation as the worked example, because the move repeats everywhere these laws are sold.

From a minimum age to age 16

The 89% is real. The manipulation is in the distance between the question it answers and the policy it is used to sell, & in who was allowed to answer the questions that came after. Parents were asked whether they want a legal minimum age at all. The policy sets that age at 16 & bans everyone younger. Between a minimum age and age 16 sits every parent who would have set the line at 13, 14, or 15, & the questions that named 16 were put only to the parents who had already said yes, with no answer left for none at all. A broad, fair number ends up standing in for a narrow, contested one. And in the one venue where people could object without a form deciding their options, a petition, far more of them did than the 8,443 the government quoted as consent.

The Biometric Update post claiming 85% of responding parents backed a legal minimum age and that the debate has shifted to how age assurance should be governed and enforced, a figure that appears in no primary source.

UK consultation: Growing up in the online world

Main article: Online Safety Act

Where the 116,211 responses came from

The consultation was called Growing up in the online world: a national conversation, run by the Department for Science, Innovation and Technology (DSIT) from March 2, 2026 to May 26, 2026. DSIT's June 2026 summary describes more than one instrument.

Respondents could submit their views through the 'full consultation' questionnaire, which included all questions, or separate, focused consultation questionnaires for children and young people up to 21, as well as their parents and carers.

The counts are large. DSIT reported 116,211 responses across all options, including 24,534 to the full consultation questionnaire, 279 unique emails, & 33,141 campaign emails, plus 39,116 responses to the parents' consultation, 5,011 to the parents' panel survey, 5,113 to the children & young people's consultation, & 9,017 to the children & young people's panel survey. Which instrument a figure comes from matters, because the famous numbers come from the full-length version, not the streamlined parent form.

Questions 4, 5, and 6 routed opponents out

Question 4 on the full consultation asked, verbatim, Would you support a legal requirement for social media services to have a minimum age of access? with options Yes, No, & Don't know/Prefer not to answer. Of 21,828 who answered, 71% (15,552) said Yes, 26% (5,671) said No, & 3% (605) chose Don't know/Prefer not. A survey that offers a real No, & logs a quarter of people choosing it, asked a fair question. Nothing is wrong yet. The manipulation is in what came next.

Question 5 then asked respondents to agree or disagree that Social media services should have a minimum age of access of at least 16 and should not be accessible to any children under that age. The instrument states who saw it.

(This question was only asked of those who responded "Yes" or "Don't know" to the question above)

The 26% who said No to Question 4 were never shown Question 5. So the agreement figures on Question 5 (78% strongly agree, 12% somewhat agree) describe a pool from which the opponents had already been removed. You cannot disagree with a statement you were never shown.

Question 4 and Question 5 of the full consultation, with the disclosure under Question 5 that it was only asked of those who responded Yes or Don't know to Question 4, so the 26% who said No were never shown the agreement question.

Question 6 offered no zero-age option

Question 6 asked Would you support a legal requirement for social media services to have a minimum age of access lower than 16? If so, at what age would you set it? It branched by how the person had answered Question 5.

(This question was asked with slight variations for those who responded 'Strongly agree' or 'Somewhat agree' to the question above, as opposed to all other responses)

Look at the variant shown to the disagree-and-other group. The options were Yes-13 (53%, 673), Yes-14 (18%), Yes-15 (4%), Don't know/Prefer not (3%), & Other (please specify) (23%, 287). There is no no minimum age at all box. A person who thinks there should be no legal age requirement is handed a menu that starts at 13 & tops out at Other. The 23% who fell into Other on that variant are the people the form had no answer for. The agree-group variant adds a No, not lower than 16 option at 81% but still presupposes a minimum age exists.

The Question 6 variant shown to respondents who did not agree, where the menu runs from age 13 to an Other box and offers no option for no minimum age at all.

89% want a minimum age, which is not the policy

The 89% comes from the parents-and-carers transparency release, published June 1, 2026. Question 12 put to parents what Question 4 put to everyone, verbatim: Would you support a legal requirement for social media services to have a minimum age of access?, answered Yes, No, or Don't know/Prefer not to answer. Of the 9,499 parents who answered it on the full-length form, 8,443 said Yes. That is the 89%.

The question is fair, it has a real No box, & 89% is a real result; most parents want some legal minimum age. The trouble is what it does not ask. It does not ask about 16. A minimum age of access could be 13, 14, or 15, the ages the consultation's own follow-up offered. A parent who would set the line at 13 says Yes to Question 12 exactly like a parent who wants 16, so the 89% folds both into one support figure for a policy that picks 16 and bans the rest.

The figure that speaks to 16 is a different one, and it is the filtered one. Question 13 asked parents to agree or disagree that the minimum should be at least 16, and the release states who was allowed to answer.

Question 13 was only presented to respondents who answered "Yes" to Question 12 (i.e. those who supported a legal requirement for a minimum age of access in principle). The 96% figure therefore relates to the level of agreement with a minimum age of at least 16 among those parents and carers who opted to respond to this Chapter and already supported some form of minimum age requirement. It does not represent the views of all consultation respondents, nor all parents and carers who responded.

So the much-quoted 96% is agreement with a 16 line among people who had already said they wanted a minimum age. It cannot tell you how many parents back 16 overall, because everyone who wanted no minimum age, or a lower one, was filtered out before the question was asked. And the gap is real: on the full consultation, even among respondents who agreed the minimum should be at least 16, 12% would have set it lower, at 13, 14, or 15; among those who did not agree, most picked 13. Wanting a minimum age & wanting it at 16 are different positions, & only the first is in the 89%.

The parental support release states that Question 13 was only presented to respondents who answered Yes to Question 12, so the 96% agreement figure does not represent all parents and carers who responded.

Simplified parent survey: no Yes or No question

That 89% came from the parents who filled in the full-length form, the one that opens with the fair Yes/No gateway. It was not the only parent survey. The government also ran a simplified version for parents and carers, hosted by Savanta, & the open version of it drew 39,116 responses, far more than the 9,499 parents who completed the full form. Savanta's published report sets out that survey's minimum-age questions, & a Yes or No on whether to have a minimum age at all is not among them. The first one already assumes the answer, asking parents only to agree or disagree that

Social media platforms should have a minimum age of access of at least 16 and should not be accessible to any children under that age.

The follow-up keeps the floor in place & only lets a parent move it down:

Would you support it being a legal requirement for social media platforms to have a minimum age lower than 16? If so, what age should it be?

The offered answers ran from 12 or younger up to 15, or No, it should be 16. A parent could disagree that the age should be at least 16, but nowhere in this survey was there a way to say there should be no legal minimum age at all. That is the distinction that matters. The full consultation asked the fair question, heard a quarter of people answer No, then routed those people out of the questions that set the age. The simplified survey, the one 39,116 parents actually filled in, never asked the question. One buries the dissent. The other never lets it into the room.

Caveats DSIT placed beside the numbers

The people who wrote the summary noticed the missing option themselves. On the free-text questions about location sharing & talking to strangers, they recorded that some respondents tried to write in a no-restriction answer & that the form could not capture it.

A very few referenced "Any age", "All ages" or similar terms; however, given the framing, it was not possible to determine whether these respondents intended to advocate for a total ban on the functionality or for no age restriction at all.

That is the design flaw stated by the authors: the framing could not tell a total-ban answer apart from a no-restriction answer. The government's own caveats go further. The summary says a consultation is a self-selecting sample, that consultations are not representative polls and are designed to elicit the range of insights on the policy, rather than assess the prevalence of opinions, & on data quality it says plainly:

A consultation is intended to collect the range of views and insights into proposed government policy. It does not measure levels of public support for a policy option.

The Data quality section of the consultation summary states that a consultation does not measure levels of public support for a policy option.

The parents release repeats the point, noting the results reflect the views of parents and carers who were motivated to take part, and are not representative of parents and carers nationally & that the figures should be treated as provisional. These are caveats about a self-selected consultation, not proof the 89% is false. They mean the document was never built to be read as a public mandate, which is how it was used.

Petitions opponents could sign

The consultation's questions about 16 never gave an opponent a clean box to tick. A petition does. Petition 757233, titled Do not ban social media for under 16s, is the box the form withheld, & it is open: 105,902 people had signed it as of June 15, 2026, & the count is live & still rising. It crossed the 10,000-signature threshold that triggers a government response on February 18, 2026, the government responded on March 4, 2026, & it crossed the 100,000 threshold for debate consideration on June 15, 2026. Those thresholds are the published rules of the system: 10,000 signatures gets a government response, 100,000 gets the petition considered for a debate in Parliament.

A petition is self-selected, & the obvious objection is correct. It is signed by people who feel strongly enough to go find it & put their name down, exactly the way a consultation response is filled in by people motivated to take part. It is not a representative poll, & I am not claiming the 105,902 prove the public opposes the ban. That is not the point. The point is the asymmetry of scale. The government amplified 8,443 parents who said Yes in a self-selected consultation & sold it as 89% support for the ban; on the cleanest oppose box the public had, more than twelve times that many people signed against this specific under-16 ban, on the same policy. When opponents were finally handed a real way to register opposition, far more of them used it than the count the government quoted as consent.

The government's own response to the petition gives away the rest. After thanking the signers, it concedes, in its own words, that opinions on a possible social media ban are divided. That is the same government that cited the 89% as the headline for the ban, admitting on the petition page that the question was contested. A consensus & a divided public cannot both be true of the same set of people; the 8,443 figure was never the first.

A second petition shows the scale of objection to the machinery these checks run on, & it has to be kept in its own bucket. Petition 730194, titled Do not introduce Digital ID cards, is closed, so its number is final: 2,984,191 signatures, debated in Parliament on December 8, 2025. The Register placed it short of the 6 million-plus petition to reverse Brexit. That nearly 3 million is not opposition to the under-16 ban, & it is not opposition to age verification as such. It is objection to a national digital-identity system, the verification infrastructure age checks increasingly run on. I keep the two petitions apart on purpose: the 105,902 is about the ban, the 2,984,191 is about the rails. The EFF kept them apart too, treating digital ID on its own terms in its December 2025 writing rather than folding it into age verification. A reader should not infer that 3 million people opposed the social-media ban; they opposed mandatory digital identity, which is a different fight that the age-check debate keeps trying to borrow numbers from.

Biometric Update's 85% figure

Biometric Update, a trade outlet that covers the age-verification industry, posted from its official account a promotion for its own article.

UK consultation results show overwhelming support for social media age limits, with 85% of responding parents backing a legal minimum age requirement. The debate is increasingly shifting from whether age assurance works to how it should be governed and enforced.

The 85% in that tweet appears in no primary source. Both GOV.UK documents report 89%, & so does the body of the same Biometric Update article the tweet is promoting, which states that 89 percent of parents and carers who responded backed a minimum age. So the headline figure broadcast to the public contradicts the document under it. The second sentence of the tweet does the heavier lifting: it reframes a contested, self-selected, routed consultation as a debate that has shifted to governance and enforcement, meaning the question of whether to do this at all is treated as closed. A routed instrument produces an inflated agreement number, the number is restated as overwhelming consent, & consent is restated as a settled mandate to move on to implementation.

Opposition recorded in the DSIT summary

The same DSIT summary that produced the 89% records the opposition, in its own words, where the headline does not go. On the substance, it reports that many respondents, particularly from industry, opposed a blanket minimum age of access, instead arguing for either a risk-based or age-differentiated access, & that this view is shared by several civil society and academic respondents. It records that other respondents rejected an Australian-style ban, citing early evidence of ineffectiveness. It records a warning that the policy could backfire: a civil society respondent stressed that without effective age assurance, any ban risks providing false assurance while children continue to access services in practice.

The numbers carry the same buried dissent. On Question 31, which asked whether adults should complete age checks more often, if it means children are safer online, 53% strongly agreed but 30% (6,513) strongly disagreed. Nearly a third rejected the premise even when it was tied to child safety. On Question 14, asking whether restricting children's access would provide a safer experience, 63% strongly agreed & 15% (3,277) strongly disagreed. And on the age-assurance questions, the summary records that the largest emerging theme was that mandatory age assurance is expected to increase personal data collection, risk privacy breaches, erode anonymity, enable surveillance, and threaten civil liberties, especially for vulnerable users. None of that is in the 89% headline.

Repeats in other jurisdictions

This is not a one-off. The shape of it shows up across jurisdictions whenever an Age verification or social-media age law needs public backing.

KOSA in the United States

In the United States, the Kids Online Safety Act was promoted with a poll claiming 86% of voters support the bill, from an online survey of 1,200 voters in late October 2023, published by Issue One. An oppose option existed, but the question described the bill purely as a tool against catastrophic harm, with no mention of its age-verification mechanics, privacy requirements, or First Amendment questions. The preamble read, in part:

As you may know, the Kids Online Safety Act, or KOSA, would require social media platforms to protect minors from online harms, such as the promotion of eating disorders, suicide, substance abuse, and sexual exploitation.

The question went on to ask whether the respondent supported or opposed the bill. When the only framing offered is stopping suicide & sexual exploitation, opposing becomes a vote to tolerate those harms. That is the social-desirability bind doing the work the missing-option does in the UK form.

YouGov: abstract rules versus login friction

YouGov found in April 2026 that 76% of Britons support banning under-16s from social media accounts. The pollster itself flagged the structural problem with that kind of figure.

In criticism of previous YouGov UK research on the Online Safety Act, critics argued that asking about support for age verification rules inflated support relative to asking about the practical implications, namely having to verify your age upon login.

Biometric Update made the same point about the label, writing that YouGov uses the most common consumer label, 'age verification,' to represent a set of technologies like facial age estimation that do not involve verification. Support for an abstract rule is not support for scanning your face to log in.

Tony Blair Institute digital-ID priming

Digital ID is the infrastructure these checks run on, & the priming version of the trick shows up there too. A Tony Blair Institute report in September 2025 claimed that Britons prefer government [83%], not the private sector [17%], to run multiple elements of a digital-ID system. The Age Verification Providers Association, the trade body for the verification industry, took apart that 83% in written evidence to the Home Affairs Select Committee, pointing at the two questions placed right before it.

But then we read the two preceding questions in the survey and it became clear how they achieved this unlikely result. 'Some are suggesting the government should introduce a new app, allowing instant access to a range of public services...'

Prime someone with a convenient government app, then ask who should run digital ID, & you manufacture the consent for state control. The notable part is who caught it: the age-verification industry's own body.

Australia's Age Assurance Technology Trial

In Australia, the 2024 under-16 social media ban leaned on a government Age Assurance Technology Trial whose preliminary 2025 findings were broadcast as age assurance can be done in Australia privately, robustly and effectively. Members of the trial's own advisory board said that overstated what the trial tested. John Pane of Electronic Frontiers Australia called the findings strong on hype, rhetoric and difficult to reconcile with the evidence, & Tim Levy of Qoria resigned from the board over the gap between the early results & the published claims. The design limit was that the trial only had to show feasibility.

The project was only ever intended as a survey to determine feasibility...the trial need only have determined that a single participant's technology could be deployed; there is no prescriptive intent as to how many participants need to have demonstrated this capability...

A study built to answer is this possible in principle was reported as proof the technology was ready to deploy on every account in the country.

Pew and YouGov: fair questions, majority yes

Fairly worded polls find majority support anyway. Pew Research Center surveyed US adults & teens in fall 2023 with a balanced item that let people pick Support, Oppose, or Not sure, & found 71% of US adults & 56% of US teens favored requiring age verification, with 16% of teens opposing. Independent YouGov polling pointed the same way, with 77% of Australians supporting the under-16 ban. Pew handed people a clear way to say no, with Support, Oppose, & Not sure all on offer, & a majority still said yes.

If the genuine number is a clear majority, you do not need a routed instrument with no opt-out to produce one. The manufactured supermajority is only useful for one thing: foreclosing the debate by claiming the public has already settled it.

What I take from this

The 89% is an honest count of a fair question: do you want a legal minimum age. Most parents do. What got built on top of it is the problem. The question never named 16; the only figure that did was shown only to people who had already agreed; the simplified survey 39,116 parents filled in never asked whether there should be a minimum age at all; the self-selected consultation was handed to the public as a mandate for a specific ban; & a trade outlet restated it as 85%. The 30% who strongly disagreed that adults should age-check more, the respondents who wanted risk-based or lower-age rules, the warning about false assurance, the privacy theme that was the largest one raised, & the far larger numbers who registered opposition the moment they had a clean way to do it: all of it is in the record & none of it is in the headline. I would rather lose an argument to a fair question than win one by quoting the right number for the wrong policy. When support has to be assembled from a vaguer question & a filtered one, that is the strongest evidence the real answer was contested.