The most damaging document in the case that opened against Meta this week in Oakland is not a leaked memo or a deleted message. It is a piece of routine growth research, competently done, that a normal company would file under standard analytics.
It is titled “Long Term Retention: The Young Ones Are The Best Ones.” It examined tweens, roughly ages 10 to 12, and it concluded what the title says: the earlier a user starts on the platform, the longer they stay and the more revenue they generate over their life as a user.
Attorneys general from 29 states are arguing that this study is not an artifact of the business model. It is the business model.
What Is Actually on Trial
Lawyers for California, Colorado, Kentucky and New Jersey delivered opening statements Tuesday before an eight-person jury in federal court in Oakland, leading a bipartisan coalition that filed in 2023. California Deputy Attorney General Megan O’Neill compressed the states’ theory into four verbs, telling jurors that Meta’s model is to “hook” users, “hold” them as long as possible, “harvest” their data, and “hide” the truth when making public statements.
NPR, covering the openings, reported the states’ framing bluntly: profits won. The claims run to deliberate design for compulsive use by minors, public misrepresentation of known harms, and collection of children’s data in violation of the Children’s Online Privacy Protection Act.
Judge Yvonne Gonzalez Rogers will decide liability herself; the jury’s verdict is advisory. If she finds against Meta she can impose civil penalties and order changes to how Facebook and Instagram operate. CNN reports the theoretical exposure at up to $1.4 trillion, a number worth treating carefully: it is statutory per-violation math multiplied across an enormous user base, not a figure anyone expects to be assessed. Meta’s market capitalization does not support it and no court would try.
The remedy that should worry Meta’s investors is the other one. A judge ordering product changes to recommendation systems, default settings, or how minors are onboarded would reach the compounding engine of the business in a way a fine never does.
Why the Study Is So Hard to Explain Away
Meta’s defense is that the states’ claims are unsubstantiated, and that its record on teen protection is strong. The company points to enhanced privacy defaults and a one-hour Instagram timer. Those are real features and they are not nothing.
The trouble is that the retention study does not allege wrongdoing, which is exactly what makes it useful to the states. It is a rational document. An advertising-funded network is valued on the lifetime value of its users, lifetime value is a function of how long someone stays, and the research establishes that acquiring a user at 11 rather than 16 extends that curve materially. Any competent growth team would run that analysis. Any competent finance team would act on it.
The states are not arguing that Meta behaved aberrantly. They are arguing that the ordinary arithmetic of an ad-supported network makes children its highest-return acquisition channel, and that Meta did the math and followed it.
That is a harder case to defend than a rogue-employee story, because there is no rogue employee to disown. It also means the discovery record is likely full of similar material, produced by people who had no idea they were writing evidence. The states have said they will present internal documents alongside testimony from seven former employees.
The first of them has already taken the stand. Arturo Béjar, a former Facebook engineering director turned whistleblower, testified that the company prioritized shipping quickly over safety. NBC News reports that Mark Zuckerberg sits on the states’ witness list for roughly three hours of testimony during a trial expected to run six or seven weeks.
The Industry Already Priced This In
The most telling evidence that this case matters is not in the courtroom. It is in what Meta’s competitors shipped in the weeks before it started.
OpenAI released a dedicated teen version of ChatGPT with age prediction built in, a product decision we read at the time as litigation risk arriving before regulation did. Nobody builds age-inference infrastructure because users asked for it. They build it because the discovery process in somebody else’s lawsuit has made the cost of not having it legible to a CFO.
That is the actual mechanism by which this trial reshapes the sector, and it works whether or not Gonzalez Rogers finds liability. Every platform with minors on it is now watching seven former employees walk a jury through internal growth research, and every general counsel in the industry is calculating what their own retention decks would look like read aloud in Oakland.
What Changes for the Business
Meta’s near-term earnings do not hinge on this. The company has absorbed regulatory losses before and its ad machine has been indifferent to most of them.
The medium-term risk is compositional. If a court forces changes to how minors are acquired and retained, the effect lands on the top of the funnel, and top-of-funnel effects in a network business show up years later as a cohort that never formed. That is invisible in a quarterly print and severe in a five-year model. It is also the specific harm the states are describing from the other direction: they say Meta understood the value of an 11-year-old precisely because it understood cohort math.
Zuckerberg has spent this year arguing that Meta’s future runs through AI rather than social feeds, a case he made at length in his manifesto on open weights and model access. Whatever else that repositioning accomplishes, it moves the story away from a set of products whose growth history is about to be read into the record for six weeks.
The Verdict Is Not the Point
There is a version of this that ends with Meta paying a penalty it can afford, appealing the injunctive relief, and continuing largely unchanged. That version is plausible and it is probably the base case.
But the states have done something more durable than build a damages claim. They have taken the internal logic of an attention business, the part that lives in growth reviews and retention dashboards and is never spoken in public, and they are putting it in front of a federal judge in the company’s own words.
“The young ones are the best ones” was written as an observation about retention curves. It will be quoted for years as something else.