On Thursday, May 29, the California Assembly voted unanimously to pass AB 1709, a bill that would ban users under 16 from social media platforms that employ “addictive features”—infinite scroll, autoplay, push notifications. The vote was 72-0. The bill gives companies until 2028 to comply. Governor Newsom is expected to sign it.

There is, on the surface, something satisfying about lawmakers finally doing something about teenage screen time. Assemblymember Josh Lowenthal, the Long Beach Democrat who introduced the bill, told colleagues before the vote that millions of California children “are spending more time on social media than they are in school.” The numbers back him up: a 2025 Pew survey found that 46% of U.S. teens say they are online “almost constantly.” The impulse to intervene is not unreasonable.

But AB 1709 has a quiet problem, and it is not the one you have been hearing about. The predictable objections—government overreach, parental responsibility, the impossibility of age verification—have all been aired. What has not been aired is that the bill targets the wrong thing entirely.

The Scroll Is the Spoon, Not the Meal

Infinite scroll is a user-interface pattern. It was invented in 2006 by Aza Raskin, a designer who has since publicly regretted it, calling it a mechanism that “keeps you in the loop.” The idea is simple: instead of clicking “next page,” new content loads automatically as you approach the bottom of the screen. It removes friction. It keeps the feed going.

But infinite scroll does not decide what appears in the feed. That job belongs to the recommendation algorithm—a machine-learning system that ranks content based on predicted engagement. The algorithm learns that outrage keeps you watching. It learns that a certain genre of video, served at a certain time of day, triggers a cortisol spike that brings you back tomorrow. The scroll is just the conveyor belt. The algorithm is the factory deciding what goes on it.

AB 1709 defines an “addictive feature” as a “psychologically exploitative feature intended to maximize engagement that foreseeably leads to compulsive use.” The bill’s text explicitly names infinite scroll and autoplay. It does not name collaborative filtering, engagement prediction models, or content ranking systems. A platform could comply with the letter of the law by replacing infinite scroll with a “load more” button—and change nothing about the algorithmic machinery that made the platform addictive in the first place.

One product designer at a mid-sized social app put it to me in a Slack DM on Thursday afternoon: “We can ship a paginated feed in a sprint. Rewriting the recsys to not optimize for dwell time? That’s the whole business.”

What the Vote Actually Protects

A unanimous vote is a strange thing in American politics. It usually means one of two things: either the bill is genuinely uncontroversial, or nobody wants to be seen voting against it. AB 1709 falls into the second category. Voting against a bill framed as protecting children from Big Tech is political poison, even if the bill’s mechanics are hollow.

The result is legislation that gives everyone cover. Lawmakers get to say they stood up to Meta and TikTok. Parents get to feel like Sacramento is on the case. And the platforms get a compliance deadline three years away—an eternity in software—to make cosmetic changes while the underlying engagement economics hum along undisturbed.

This is not a new pattern. In 2022, California passed the Age-Appropriate Design Code, which required platforms to consider children’s best interests in product design. A federal judge blocked key provisions in 2023, and the law has been tied up in litigation ever since. AB 1709 may meet a similar fate, but even if it survives, the compliance path is narrow enough to drive a truck through.

The Harder Conversation Nobody Wants

If California were serious about algorithmic harm, it would have to do something genuinely uncomfortable: regulate the recommendation engines themselves. That would mean auditing the models. It would mean requiring platforms to demonstrate that their content-ranking systems do not systematically amplify content that harms minors. It would mean transparency into the objective functions—the mathematical targets—that these systems are trained to maximize.

That is a much harder bill to write. It is also a much harder bill to vote for, because it threatens the core revenue model of some of the most valuable companies in the state. Meta generated $165 billion in revenue in 2025, nearly all of it from advertising sold against engagement. An algorithm that optimizes for something other than engagement is an algorithm that generates less revenue. Lawmakers know this. The platforms know this. The bill’s silence on the matter is not an oversight.

AB 1709 will probably become law. Platforms will redesign their feeds, add some friction, and issue press releases about their commitment to teen well-being. The infinite scroll may indeed become endangered. But the algorithm that made the scroll worth scrolling will remain untouched—because banning a UI pattern is easy, and banning a business model is not.

None of this means the bill is worthless. A “load more” button is marginally better than an endless feed, if only because it introduces a moment of conscious choice. But anyone who believes AB 1709 will meaningfully reduce teenage screen time should ask themselves: when was the last time a “load more” button stopped you from loading more?

Sources