The recent New York Times reporting on the California lawsuit against Meta and YouTube should make anyone in technology pause, not because it is sensational, but because it is plausible.

In opening arguments, attorney Mark Lanier stacked three children’s blocks in front of a Los Angeles jury and labeled them “Addicting. Brains. Children”, as easy as ABC. It was a visual designed to land. Whether it ultimately persuades a jury is another matter. But the symbolism cuts close to something many of us in tech understand, even if we don’t like to say it out loud.

If the internal documents described in the reporting are accurate, this case isn’t about rogue content slipping through moderation filters. It isn’t even primarily about speech. It’s about design. More specifically, it’s about what happens when engagement becomes the dominant product objective and children become a high-performing cohort inside that model.

According to the reporting, internal materials allegedly compared certain product features to slot machines. Another document highlighted that users who joined at 11 had significantly higher long-term retention than those who joined at 20. Anyone who has ever sat in a product review meeting knows exactly what those numbers mean. Retention curves aren’t philosophical. They’re operational. Higher lifetime value changes strategy. It influences roadmap priorities. It reshapes acquisition focus.

None of that is inherently sinister. Cohort analysis and retention modeling are standard practices in digital businesses. But context matters. When the users in question are minors, and the system is engineered to maximize time on the platform, the moral neutrality disappears quickly.

We tend to use the word “engagement” as if it’s synonymous with “participation.” In ad-driven models, it’s not. Engagement is revenue. The longer a user stays, the more ads they see. The more they interact, the more data improves targeting. The more personalized the feed becomes, the harder it is to leave.

Infinite scroll reduces friction. Autoplay removes exit points. Recommendation engines escalate content intensity to sustain attention. None of these are accidental outcomes. They are refinements, tested and optimized over time. That is what modern product teams are trained to do.

The legal debate will revolve around causation. Can a specific platform be proven to have caused a specific individual’s mental health harm? That is a high and complex bar. Human psychology does not operate in single-variable models.

But from a design perspective, the more relevant question is simpler: Were these systems deliberately optimized to increase compulsive use, and were younger users identified internally as especially valuable?

If the answer is yes, then we are no longer talking about unintended consequences. We are talking about incentive alignment.

Meta’s reported defense suggests that the plaintiff’s struggles stem from family circumstances rather than Instagram. That may be partially true. Personal environments matter. But the argument frames the issue as binary, either the platform caused harm or external factors did.

That framing misses the structural point.

If a teenager is already vulnerable, and the system is designed to detect and amplify emotional triggers to keep users engaged, vulnerability doesn’t disappear inside that system. It becomes predictable. And predictability, in optimization terms, is valuable.

This is where the parallels to Big Tobacco become uncomfortable. Internal awareness of risk. Youth targeting. Public assurances alongside private modeling. The difference today is technological sophistication. Cigarettes did not adapt in real time. Algorithms do. Machine learning systems test variations continuously, refining engagement triggers at scale. They are extraordinarily effective.

In enterprise settings, we don’t deploy powerful AI systems without governance layers. We talk about oversight, risk mitigation and responsible use. We design audit trails and review boards. We debate ethical frameworks.

Yet when optimization engines operate in consumer social platforms aimed at minors, the dominant governance mechanism has historically been growth.

That is not a moral accusation. It is an economic reality. If revenue scales with time-on-platform and younger users demonstrate higher lifetime value, the incentive to attract and retain them becomes structural. Engineers respond to KPIs. Product managers respond to retention dashboards. Executives respond to growth metrics. Over time, those incentives produce predictable outcomes.

This is why Section 230, while important, may not be the central issue here. The legal shield protecting platforms from liability for user speech does not automatically resolve questions about deliberate behavioral design. These lawsuits, as described in the New York Times, focus less on what users posted and more on how the systems were architected.

That distinction matters. We are not debating the policing of speech. We are debating the engineering of reinforcement loops.

And this is bigger than one case. States, school districts and regulators globally are circling similar concerns. Australia has imposed age-related restrictions. European regulators are tightening youth protections. In the U.S., attorneys general are increasingly framing social media not just as a communications platform but as a public health issue.

None of this means technology is inherently harmful. I have spent my career championing innovation and building media businesses around digital ecosystems. Social platforms have created enormous value — connection, community, creativity, access.

But innovation without restraint eventually collides with its own incentives.

There is a meaningful difference between building products that people choose to use and engineering systems that are exceptionally difficult to disengage from — particularly for users whose impulse control and cognitive development are still forming.

The defense in these cases will focus on legal standards, causation thresholds and scientific ambiguity. That is their job. Outside the courtroom, the industry faces a more strategic reckoning.

If our most advanced optimization systems disproportionately succeed with the users least equipped to resist them, what does responsible product design require?

That is not an anti-technology question. It is a governance question.

Building a digital casino is an engineering accomplishment. Designing the house to maximize play is a business decision. Deciding who gets invited inside, and how young they are, is something else entirely.

And eventually, those distinctions become impossible to ignore. You don’t get to build the casino, rig the games for children and then claim you’re not responsible when they can’t walk away.