A judge just wrote the rules for kids on Instagram. That should trouble you even if you think Meta earned it.

The popular reading of last week's ruling out of New Mexico is that a court finally made Meta pay. On August 6, Judge Bryan Biedscheid of the First Judicial District Court in Santa Fe issued a 68-page decision ordering the company to fund a $567 million abatement program for the harm its platforms have done to the state's young people. Stacked on top of a $375 million jury verdict from earlier in the trial, that brings the total to $942 million. The judge found that Facebook and Instagram constitute a public nuisance in New Mexico, and he rejected Meta's argument that Section 230 immunised it. Attorney General Raúl Torrez called it accountability. For a lot of people who have watched the political system fail to touch this problem for a decade, it reads as a straightforward win.

I want to make an unpopular argument, and I want to be careful about what it is and isn't. It is not that the harms are fake, or that Meta is sympathetic, or that the company shouldn't answer for what a jury found. The harms are real, the jury found genuine violations, and I have no brief for the defendant. My argument is narrower and, I think, more durable: governing the design of a global software product through a public-nuisance verdict, one courtroom and one state at a time, is the wrong instrument for the job — and the reasons are procedural, not sentimental. I say this as someone who spent years inside a regulator watching good intentions produce rules that couldn't be enforced.

What the court actually did

Read past the dollar figure and look at the order itself, because the money is the least interesting part. The injunction tells Meta how to build its product for minors in New Mexico. It must hide public Like counts from users under 18 unless a parent or guardian approves. It must pause push notifications to those users between 10 p.m. and 7 a.m. It must cap their usage at 90 hours a month — roughly three hours a day. The $567 million fund is structured like the opioid settlements that inspired it: a reported $420 million for treatment services for young people, the rest for awareness, prevention and screening, spread over five years.

Set aside whether those are good product rules. Some of them may be. Notice instead what just happened: a state trial court wrote them. A notification curfew, a monthly time budget, a specific UI change gated on parental consent — these are product-design regulations, the granular kind that ordinarily come from a statute or an agency rulemaking after notice, comment, and a hearing record. Here they came from a single judge, applying a doctrine built for a very different kind of problem, at the end of a trial. That is the thing to examine, and cheering the defendant's discomfort is a way of not examining it.

The strongest case for cheering, stated fairly

It deserves to be stated fairly, because it is strong. The political branches genuinely abdicated here. Congress has spent years failing to pass child-safety legislation; the bills die in committee or on the floor, session after session. Section 230 has, for most of the internet era, blocked exactly these suits before they reached a jury, which is why so little has ever been tested on the merits. Into that vacuum, a New Mexico jury found that Meta committed some 75,000 violations of the state's Unfair Practices Act and imposed the maximum penalty the law allowed. This was not a judge free-styling; it was the back half of a two-phase trial with a factual record.

And the abatement model has a track record. The opioid settlements, built on the same public-nuisance theory, moved real money to real treatment when every other mechanism had stalled. When the legislature won't act and an agency can't reach the conduct, litigation is the system's backstop, not a betrayal of it. That is the honest version of the case for last week's ruling, and anyone who wants to argue the other side — as I am about to — has to carry it, not caricature it. I will concede the whole of it. The harms are documented, the other doors were locked, and the court used the one that was open.

But a judgment is not a regulator

Here is where the procedural reality bites. A regulation is not just a rule; it is a rule plus a body that can update it, monitor it, and enforce it over time. A verdict is a rule frozen at the moment it was written, supervised by a court that was never built to supervise a product. Start with updating. Ninety hours a month is a number a judge chose. Software changes weekly. When Meta reworks its teen experience — which it will, repeatedly, some of it in good faith — who re-tunes that cap? Who decides whether the new design satisfies the curfew, or evades it? A statute delegates that to an agency with engineers on staff. A judgment delegates it to the next round of litigation. You do not govern a moving target with a fixed order; you either freeze the product in place or you relitigate it, and neither is regulation.

Then enforcement, which is the question I always come back to: who actually watches? A court has no telemetry, no continuous audit, no ability to see what Meta ships to a 15-year-old in Albuquerque next spring. Verifying that push notifications really stopped at 10 p.m., that the usage cap is real and not cosmetic, that Like counts are hidden by default and not by dark-patterned exception — that is an ongoing supervisory function, and the court that issued the order cannot perform it. In practice it falls to the Attorney General's office, which now becomes a de facto product regulator for one company, without the staff, the standards, or the machinery a real regulator would have. "Meta must limit usage in New Mexico" is a sentence a judge can write and no courtroom can monitor.

A regulation is a rule plus a body that can update and enforce it. A verdict is a rule frozen the day it was written, watched by a court that was never built to watch a product. — On why a judgment isn't a regulatory regime

Fifty courtrooms, fifty product specs

Now multiply it. New Mexico's order governs the product "in the state." There is nothing special about New Mexico's version of the public-nuisance doctrine; every state has one, and other attorneys general are running the same play. Picture the endpoint honestly: different judges in different states, on the same elastic theory, ordering different and possibly incompatible product rules — a notification curfew here, a different time cap there, a distinct consent flow somewhere else. Design-by-litigation produces a patchwork, and a patchwork is a compliance burden that scales with legal complexity.

Ask the question I always ask: who can actually absorb that? The answer is the largest firms — the ones with the lawyers and the engineering slack to ship fifty regional variants of a teen account. A regime that can only really be complied with by a company the size of Meta is a regime that entrenches Meta. That is close to the opposite of what the people cheering this outcome want, and it is the predictable result of setting product rules through fifty separate courtrooms instead of one legislature. The instrument selects for the incumbent it was aimed at.

The appeal is the tell

There is one more reason to be wary of treating this as settled policy: it isn't settled anything. Torrez himself said the appeal could take "months, if not years." The two loadbearing legal moves here — stretching public nuisance to cover the design of a software service, and getting past Section 230 to reach the product decisions underneath it — are genuinely unsettled, and they are exactly the kind of questions appellate courts exist to answer. The remedy announced with a press conference is provisional. A higher court may narrow the nuisance theory, or restore some of the Section 230 shield, or leave the fund and trim the injunction. Building the country's de facto teen-safety rules on a doctrine that is mid-appeal is building on a foundation that may be under renovation. That is not a reason to do nothing; it is a reason not to mistake a trial-court order for a durable rule.

The better frame, and what it costs

So what is the realistic version? It is the unglamorous one: legislate the duty. Write an age-appropriate design code that states the specific obligations platforms owe minors — defaults, disclosures, the treatment of engagement features aimed at kids — with defined standards rather than a judge's improvised numbers. Name a regulator with the technical staff to audit compliance and the authority to update the rules as products change. Set a single national floor so the obligation is uniform instead of assembled from fifty verdicts. That is a rule made by a body that can revise it, enforced by one that can monitor it, applied evenly. It is what actually governing this would look like.

And I owe you the cost of my own position, because a column that pretends its preferred answer is free is not worth reading. The cost is obvious: that legislation is precisely what has failed to pass, over and over, which means "legislate the duty" can sound like counseling patience while children are harmed now — and the families in this case did not have the luxury of waiting for a Congress that wasn't coming. There is a second cost: a national floor can preempt stronger state protections, and a weak federal standard written by a captured process could leave kids worse off than a patchwork of aggressive state courts. Those are real, and I won't wave them away. My claim is not that the New Mexico court was wrong to act with the tool it had. It is that a public-nuisance judgment is a signal flare, not a regulatory regime, and the danger is that treating it as the second lets the branches that should be writing the rule keep not writing it.

That is the harder question under the celebration. Not whether Meta harmed young people in New Mexico — a jury already said it did. The harder question is whether a public-nuisance verdict should be the thing setting the notification curfew on a teenager's phone, and what it means that this was the only branch of government able to write the rule at all. A court reached the conduct because everyone better suited to reach it declined. We should be uneasy about the reaching even as we admit the reason for it — and we should save most of the unease for the empty chairs where the rule was supposed to be made.