My Wall Street Journal essay (8/28/26) on the Meta paradox—published under the headline “The Meta Settlement Makes Big Tech Bigger”—produced what one might expect from a discussion of children, social media, government and Mark Zuckerberg: a calm seminar conducted at the volume of a fire alarm.
One reader said I should never be allowed to sit “unsupervised behind a keyboard.” Another submitted the column to an AI detector and reported the verdict as though Pangram had just returned from Mount Sinai. Zuckerberg’s yacht made an appearance. So did broken glass, dirty needles and loaded guns. This is the comment section’s traditional warm-up before the argument begins.
But the argument did begin, and much of it was worthwhile. Several readers identified the point of the column more clearly than some of its critics did. “Government: ‘Big Tech is too powerful!’ Also government: ‘Here’s a mountain of regulations only Big Tech can afford,’” one wrote. Another compared the settlement to the tobacco agreement, which created a moat around the companies it was supposed to punish. A third noticed the privacy problem: to identify children reliably, Meta will have to know considerably more about everyone.
Those readers understood my argument. Others thought I was claiming that social media is harmless, Meta is innocent, parents are omnipotent and government should do nothing. I said none of those things. The disagreement is worth sorting out because it concerns more than one company. It is about how we make public policy, who gets to make it, and whether a government campaign against corporate power can wind up increasing that power.
What I argued—and what I did not
I do not doubt that some young people are harmed by social media. Anyone who has watched a teenager disappear into a screen can see the problem. Features built to hold attention can become compulsive. Bullying, sexual exploitation, eating-disorder content and relentless social comparison are real dangers. The Surgeon General has said that social media presents a meaningful risk of harm while also providing benefits. The National Academies has been more careful about causation, concluding that effects vary greatly among users and that the evidence is often associational, reciprocal and difficult to disentangle. That uncertainty is not an excuse for indifference. It is a reason to demand policies that can be measured.
Nor did I argue that every term of the settlement is foolish. My Journal piece said explicitly that some provisions may be sound policy. Muting school-hour notifications, making parental controls easier to use, restricting cosmetic-surgery filters and offering a chronological feed all may help. A useful feature, however, does not redeem a defective governing structure. The question is not whether one can find a good idea somewhere in a 122-page agreement. The question is whether state attorneys general, bargaining behind closed doors with the dominant firm, should write an operating code for an entire communications industry.
That distinction disappeared in some of the replies. One reader said my argument was merely an “excuse for doing nothing.” Another said the settlement may be imperfect but is “better than nothing.” That is the favorite defense of bad policy because it places every burden on the critic. Yet the alternatives are not this settlement or anarchy. Legislatures can enact prospective, generally applicable rules. Regulators can enforce laws against deception, privacy violations and unlawful conduct. Courts can adjudicate claims by people who can prove injury and causation. Parents can use controls, and companies can be required to make those controls intelligible. Competition can reward services that offer families safer products. There is quite a lot between 122 pages negotiated with Meta and nothing.
Parents are neither omnipotent nor irrelevant
The most serious criticism concerned parents. Several readers said modern technology has moved beyond parental control. The platforms are sophisticated, the children know the devices better than their parents, and the products are designed to overcome self-restraint. One compared Instagram to a loaded gun. Another compared parental choice to letting children play in a park strewn with broken glass and dirty needles.
Those analogies express understandable alarm, but they assume the conclusion. Social media is not a gun, and a communications platform is not a public playground that a parent cannot alter. Instagram can expose a child to genuine risks while also providing friendship, entertainment, information and creative opportunity. The risks vary by child, content, duration and circumstance. That is precisely why parents, who know the child, cannot be written out of the equation.
They also are not written out of the settlement. Its most important restrictions are defaults that a supervising parent may relax. A parent can approve more than two hours of use, modify nighttime limits and change protective settings. The regime therefore rests on an awkward premise: parents are too powerless to govern their children’s use, yet sufficiently competent to authenticate themselves, link accounts, understand the options and decide when to override them. If parental judgment is hopeless, the settlement’s own machinery is hopeless with it.
The better answer is shared responsibility. Meta should not deceive families, conceal material risks or design features that create unreasonable hazards. Government has a legitimate role in enforcing those duties. Parents still decide whether a child gets a smartphone, which apps may be installed, whether phones enter bedrooms and what happens when rules are evaded. Saying that parents retain the first responsibility is not anarchism. It is an acknowledgment that no auditor, attorney general or algorithm will ever know a child as well as his family does.
The tobacco comparison is in the agreement
Several readers compared the deal to the 1998 tobacco settlement. One critic called that analogy “specious” because society does not need new tobacco companies or new platforms that are even better at addicting children. That response does not refute the concern about entry barriers. It embraces them.
More important, the tobacco comparison is not rhetorical decoration. It is written into the settlement’s economic machinery. The agreement guarantees the states roughly $12.1 billion and makes about $5 billion more contingent on what it calls “Industry-Wide Adoption.” It names TikTok, YouTube and Snap as “Core Industry Members.” For the tighter restrictions and contingent payments to take effect, those rivals must come under substantially equivalent obligations. The agreement even defines a “New SMP Entrant”: a competing service with at least five million monthly American teen users and an average of 30 minutes of daily teen use. If such an entrant appears without joining the regime, the condition of industry-wide adoption can fail.
In other words, the states acquire a direct financial interest in extending Meta’s bargain to Meta’s current rivals and any successful future challenger. Meta also receives a most-favored-nation provision: if a state later gives a competitor more favorable treatment, Meta can demand comparable terms. This is not a theory about regulatory capture assembled by suspicious readers. It is the architecture of the agreement.
Perhaps some people would prefer a closed club of large social-media companies to open competition. But they should say so plainly and ask Congress to make that choice. Social media is not tobacco. It is a medium of speech, association, commerce and invention as well as a source of risk. A smaller competitor might be safer, less intrusive, easier for parents to understand or built around a different business model. A policy that makes such experimentation less likely should not be smuggled into national practice through a settlement with the incumbent.
The payment’s size does not answer this objection. Several readers noted that $17 billion over a decade is manageable for a company that earned about $61 billion last year. They are right—and that strengthens the paradox. A cost that Meta can absorb may still be forbidding to the firm that might challenge it. “Pocket change” for the incumbent can be a locked gate for the entrant.
The privacy paradox is worse in the fine print
One reader made an especially sharp point: to verify parental authority, Meta may need information identifying the parents, proving guardianship and linking adults to children. The actual agreement goes further. It permits ID verification and facial-age estimation. It directs Meta to analyze posts, comments and other account activity for evidence of age; review the friend networks of accounts removed as under 13; link accounts through “soft-matching” models; and use signals such as device identifiers, phone numbers and email addresses to find secondary accounts. Meta must develop a model for predicting which users are under 13 and use it to evaluate accounts across its platforms.
The agreement includes data-minimization, deletion and security requirements, and those safeguards matter. But it also permits Meta to retain certain under-13 data to develop and test the age model. The settling attorneys general expressly promise not to sue Meta under COPPA or analogous state law for maintaining or using a child’s information for that purpose, subject to the agreement’s limits.
That is the privacy paradox in black and white. Government accused Meta of collecting too much information about children. Its remedy requires more powerful identity inference, more account matching and more scrutiny of user behavior, then grants Meta a limited legal peace for retaining some of the resulting data. The intentions may be protective. The institutional capability being built is still one that civil libertarians should examine very carefully.
No jury “proved” the states’ case
Some comments confused settlement with adjudication. “All those states sued, and they won,” one reader wrote. Another said government had “proved legally” that Meta caused mental illness in children. A third portrayed the settlement as a jury award from neighbors telling injured families, “We hear you.”
That did not happen in this case. The parties settled during trial. There was no verdict on the states’ claims and no finding that Meta caused a general youth mental-health crisis. The agreement says expressly that it is for settlement purposes only and is not an admission of liability, wrongdoing or violation of law. Meta may have settled because it feared catastrophic exposure, damaging evidence, years of litigation or all three. A settlement price tells us about litigation risk. It does not convert disputed allegations into scientific or legal findings.
The same care is required when discussing the money. Readers who assumed that every dollar will vanish into unrestricted state budgets went too far. The agreement identifies youth mental-health services, crisis lines, after-school programs, digital literacy and related remediation as permitted uses, and some states have adopted tighter restrictions. At the same time, it allows broad state-specific discretion, payments into some general funds, reimbursement of litigation and enforcement costs, and other lawful uses. It is not a direct compensation program for the families whose stories drove public support for the case. The public should track where the money goes and what it accomplishes, not merely admire the number in the headline.
Section 230 did not create every problem
One reader offered a wonderfully Washington solution: blame one statutory provision for everything. Repeal Section 230, he argued, and capitalism will take its course. Section 230 generally prevents an online service from being treated as the publisher or speaker of content supplied by someone else. It does not immunize a company from every claim about its own conduct, promises or product design. Indeed, these lawsuits proceeded precisely because courts allowed important claims based on Meta’s own alleged design and business practices to continue.
Wholesale repeal would also reproduce the competitive problem at the center of my column. If every platform faced sweeping liability for millions of user posts, Meta could build another compliance fortress. A startup could not. Reform may be warranted, but “repeal it and let capitalism work” ignores what ruinous, open-ended liability does to entry.
What a better policy would look like
One reader asked, fairly, whether my job was only to identify flaws or whether I had a better solution. Here is the beginning of one. Congress—or state legislatures acting through ordinary law—should establish clear, prospective and platform-neutral rules for services used by minors. Age assurance should disclose the minimum necessary information, preferably an age-band signal from a device or app store rather than identity documents collected anew by every platform. Parents should receive simple controls and retain the ability to impose stricter limits. Researchers should receive privacy-protected access to data so that causal claims and proposed remedies can be tested. Rules should contain measurable objectives, public reporting and sunset dates. Antitrust authorities should examine whether compliance costs or industry-wide settlements foreclose smaller competitors.
That program would not satisfy people who demand immediate punishment, and it would not please those who want government entirely absent. It would do something more useful: separate child protection from incumbent protection.
One of the best comments asked what happens if Meta complies fully and teenage depression, self-harm and bullying do not measurably improve. Will the states admit that other forces matter? Will they return the money? Of course not. The settlement does require an auditor to examine implementation and some measures of efficacy, but much of the underlying material will remain confidential, some recommendations are nonbinding, and Meta retains substantial discretion over design changes.
That makes independent public measurement essential. Over the next five and ten years, we should ask: Did the rules improve outcomes for children, or did usage simply migrate to other services? How many adults and minors were misclassified? How much additional personal information was collected or inferred? Where did the settlement money go? Did new competitors enter? Did Meta’s position strengthen? A policy sold as a historic victory should be able to survive a historical audit.
The children deserve protection. They also deserve a government capable of distinguishing protection from protectionism. Meta can survive $17 billion and 122 pages of obligations. The harder question is whether the next competitor can—and whether children will be any safer if it cannot.
That is the Meta paradox. The comment section did not refute it. In more than a few cases, it supplied the footnotes.
Sources
Meta multistate Settlement Agreement, Aug. 26, 2026: https://oag.dc.gov/sites/default/files/2026-08/Settlement%20Agreement%20FINAL%20-%20Fully%20Executed.pdf
National Academies of Sciences, Engineering, and Medicine, Social Media and Adolescent Health (2024): https://nap.nationalacademies.org/catalog/27396/social-media-and-adolescent-health
U.S. Surgeon General, Social Media and Youth Mental Health: https://www.hhs.gov/surgeongeneral/reports-and-publications/youth-mental-health/social-media/index.html
47 U.S.C. § 230: https://www.law.cornell.edu/uscode/text/47/230