Facebook is the canonical case study in platform power — not because it was the first social network, but because it was the first to demonstrate how a social media platform could become a governing institution without ever being elected. Launched in 2004 by Mark Zuckerberg at Harvard, Facebook evolved from a college networking site into a global infrastructure for social coordination, information distribution, and political mobilization, reaching over three billion monthly active users at its peak. Its significance lies not in its features but in its architecture: the social graph it constructed became the default map of human social relations for a generation, and the News Feed it introduced became the template for algorithmic curation of information at scale.
The Architecture of Social Graph Power
Facebook's core innovation was the explicit modeling of social relations as a networked data structure. Unlike earlier platforms that treated users as isolated profiles, Facebook encoded the connections between people — friendships, interests, affiliations — into a computable graph. This made possible a new kind of power: the power to infer. By observing who was connected to whom, what they engaged with, and how information propagated through the network, Facebook could predict preferences, behaviors, and vulnerabilities with increasing precision.
The social graph is not merely a database of friendships. It is an attention architecture that determines whose voices are audible and whose are drowned out. The algorithmic curation of the News Feed does not simply select content; it selects the social reality that each user inhabits. Two users with the same set of friends may see entirely different worlds, shaped by engagement optimization rather than editorial judgment. This is the mechanism behind filter bubbles and epistemic fragmentation at population scale.
The Engagement Economy in Action
Facebook perfected the engagement economy. Its business model — free to users, paid for by advertisers — required maximizing time-on-platform, which in turn required maximizing emotional engagement. The platform learned, through billions of behavioral experiments, that outrage, fear, and moral superiority produced more clicks, shares, and comments than nuance, accuracy, or calm reflection. The algorithm did not need to be designed to promote polarization; it merely needed to be designed to maximize engagement, and polarization emerged as an emergent property of the system.
The Cambridge Analytica scandal of 2018 revealed the downstream consequences: the same data infrastructure that enabled targeted advertising also enabled targeted manipulation. But the scandal missed the deeper point. The problem was not a single rogue actor exploiting the system. The problem was the system itself — a complex system whose components (profiles, likes, shares, algorithms, advertisers) interacted to produce outcomes no designer intended, including the erosion of democratic discourse, the amplification of conspiracy theories, and the collective cognition of populations increasingly unable to share a common factual baseline.
Platform Governance and the State
Facebook's scale forced a structural question that democratic theory had never adequately answered: what happens when a private corporation controls the infrastructure of public communication? The platform's decisions about content moderation, algorithmic amplification, and political advertising were effectively governance decisions — but made by unelected executives, governed by terms of service rather than constitutional law, and optimized for shareholder value rather than public good.
The renaming of Facebook to Meta in 2021 was an attempt to reframe the company as a metaverse builder rather than a social media platform. But the rebranding did not change the architecture. The same social graph, the same engagement-optimization systems, the same information cascades remained in place. The platform's power was infrastructural, and infrastructural power is harder to regulate than behavioral power. You cannot ban a harmful post and fix the system when the harm is in the algorithm itself.
The failure to regulate Facebook is not a failure of political will but a failure of regulatory imagination. We have frameworks for products and services, but Facebook is neither — it is infrastructure, and infrastructure requires a different category of governance entirely. The idea that users can 'choose' to leave Facebook misunderstands the nature of network effects: when three billion people are on a platform, leaving is not a consumer choice but a form of social exile. The platform is the public sphere now, and the public sphere cannot be governed by terms of service.