We've spent two decades building platforms that monetise human attention. This essay asks whether the structural logic of that economy can ever be reformed from within.
In 2004, a Harvard sophomore built a website that let students rate the attractiveness of their classmates. Within hours it had crashed the university's network. Twenty years later, the company that grew from that experiment is worth over a trillion dollars and has reshaped — in ways we are still struggling to articulate — the texture of daily life for billions of people.
The story of social media is, at one level, a story about attention. Not attention as the cognitive scientists mean it — the directed focus required for complex tasks — but attention as an economic resource, something that can be harvested, packaged, and sold. The attention economy is the name we give to the set of markets and incentive structures that have developed around this harvesting.
The basic model is straightforward. Platforms provide services — social connection, information, entertainment — at zero monetary cost to the user. The platform monetises through advertising, which means its business interest lies in maximising the time users spend on the platform. Every product decision, from the infinite scroll to the notification badge, serves this end.
What makes this arrangement distinctive is not its venality — plenty of industries seek to maximise consumption — but its precision. Digital platforms have access to granular data about user behaviour: what you click, how long you linger, when you return. This data allows them to optimise their products for engagement with a thoroughness that traditional media could never approach.
The problem is not that the platforms are malicious. The problem is that they are functional — they work exactly as designed.
Here is the structural problem. Engagement — the metric that advertising-funded platforms optimise for — is not the same thing as value. Research consistently finds that content which provokes strong emotional reactions, particularly outrage and anxiety, generates more engagement than content that is accurate, nuanced, or genuinely informative. A platform optimising for engagement will, all else equal, surface more outrage and more anxiety.
This is not a bug that can be patched. It is a feature of the incentive structure. As long as advertising remains the primary revenue model, the interests of platforms and the interests of users will systematically diverge.
The reform proposals are broadly of three kinds: regulatory (governments impose constraints on platform behaviour), structural (change the revenue model, e.g. subscriptions or public funding), and technical (design platforms to optimise for user wellbeing rather than engagement). Each has merits; each has significant obstacles.
Regulatory approaches face the classic problem of jurisdictional arbitrage — platforms operate globally, regulators nationally. Structural reforms require either user willingness to pay or public subsidy at significant scale. Technical reforms founder on the question of how you measure wellbeing, and who decides.
None of this means reform is impossible. But it does suggest that it will be slow, contested, and incomplete — and that in the meantime, the default will continue to be an economy of attention that serves advertisers rather than the people whose attention is being sold.