The Advertising Model and Its Externalized Costs

The behavioral harm documented in Paper I of this series — dopaminergic disruption, attention fragmentation, adolescent mental health deterioration — is not a byproduct of technology. It is the predictable output of an economic architecture that rewards maximizing user engagement, externalizes the psychiatric cost of engagement optimization, and operates in a market structure that prevents competitive self-correction.

The harm is not a bug. It is an externality — a cost borne by users while profits are captured by platforms. Standard economic analysis predicts that markets will not self-correct externalities. This one has not. This brief makes the economic case for regulatory intervention.

The advertising revenue model creates inescapable structural pressure: every product improvement from the user wellbeing perspective that reduces time-on-platform also reduces advertising revenue. The conflict between user wellbeing and platform revenue is not incidental. It is architectural. A platform cannot simultaneously maximize user engagement and user wellbeing when wellbeing requires less engagement.

The metrics that govern platform operational priorities — Daily Active Users, Monthly Active Users, time-on-platform, return rate — all point in the same direction: more is better. None of them measure whether the time was good for the user. What gets measured is what gets optimized. What does not get measured is what gets externalized.

The result is predictable and predicted by basic economic theory: the engagement optimization that maximizes these metrics — variable reward scheduling, algorithmic amplification of emotionally activating content, infinite scroll, compulsion-optimized notifications — produces psychiatric harm that is not captured in platform revenue accounts and therefore creates no financial incentive for platforms to reduce it.

An externality is a cost that falls on parties outside the transaction that produced it. Carbon emissions are the textbook case: the firm captures production benefit while the atmosphere bears the cost. The attention economy follows the same structure: platforms and advertisers capture revenue; adolescent users bear the psychiatric cost in the currency of their mental health and developmental trajectory.

The market does not self-correct externalities. This is not an ideological claim — it is one of the most foundational propositions in welfare economics, with no serious dispute in the literature. External intervention — regulation — is what internalizes externalized costs when markets cannot.

The scale of the externalized cost is difficult to quantify precisely, but its lower bound is significant: the United States spends approximately $280 billion annually on mental health treatment and services. The fraction attributable to adolescent social media exposure is contested but substantial. This is cost that is not recorded in any platform account.

Three structural features of the attention economy prevent the competitive self-correction that would normally discipline harmful products:

Network effects.

The value of a social platform depends on how many of a user’s connections are on the same platform. A competitor offering better psychological outcomes cannot offer the same network. The most beneficial platform in the world cannot displace an established platform until it has already achieved mass adoption — which it cannot achieve without already having it.

Switching costs.

Years of social history, content archives, and follower relationships are not portable. The investment in an existing platform reduces the competitive pressure that would otherwise discipline its behavior.

Data advantages.

Advertising targeting precision — the mechanism that generates revenue — depends on years of accumulated behavioral data. New entrants cannot match this regardless of product quality. The data advantage compounds and is difficult to replicate.

No healthy competitor gaining share through better user outcomes has achieved meaningful scale against engagement-optimized incumbents. The market has been running this experiment for a decade. The self-correction hypothesis is falsified by the evidence.

Tobacco companies had no structural incentive to reduce the harm of their products until the regulatory and legal environment made harm costly. They did not voluntarily develop less harmful products or reduce deceptive marketing. The incentive architecture changed. The behavior followed.

The same logic applies here. Voluntary corporate responsibility commitments do not change the optimization target. Liability exposure, advertising restrictions, and age-based access limitations change the incentive architecture. When the incentive architecture changes, behavior changes. This is the mechanism. It works. It has worked before.

The extraction machine was not an accident. It was the logical product of an advertising business model applied to behavioral data at scale. The harms are externalized because the architecture externalizes them. Changing the architecture requires changing the incentives. This is what regulation is for.