There was a time, not long ago, when a significant portion of a nation’s population watched the same television programme on the same evening, read the same newspaper over breakfast, and knew the same songs from the same radio stations. This cultural condition is now called monoculture, and its passing is almost universally described as progress. More voices, more diversity, more choice: the argument writes itself, and it is not wrong.
But it is incomplete. Something else was lost when the monoculture ended, and we have not yet found adequate language for what it was.
The monoculture was a commons. It was built on exclusion and power asymmetry, and its content reflected the prejudices of whoever controlled the distribution channels. These are real and serious criticisms. But it was also a space of genuine collective reference — a vocabulary shared across political, class, and regional lines that made certain kinds of public conversation possible. The politician who quoted from a television programme everyone had watched was building a bridge into a shared experience. The satirist who targeted a shared cultural object was able to do so because the target was shared.
People do not simply disagree about political questions. They increasingly inhabit separate informational and cultural worlds, with limited shared reference.
Algorithmic personalisation has dissolved this commons. The mechanism is well-documented. Recommendation systems, designed to maximise engagement, systematically surface content that confirms existing preferences and avoids the friction of exposure to different views. The result, documented in multiple studies from MIT’s Media Lab and the Oxford Internet Institute, is not merely political polarisation — that is the most discussed symptom — but a more fundamental fragmentation of shared cultural reality. People do not simply disagree about political questions. They increasingly inhabit separate informational and cultural worlds, with limited shared reference.
The implications for democratic institutions are more severe than the polarisation data captures. Democracy requires not agreement but the capacity for shared deliberation — the ability to argue within a shared framework of facts, norms, and cultural reference. That framework is the product of a common public sphere, and the public sphere requires some degree of shared experience. When citizens have no common cultural vocabulary, political discourse becomes not debate but mutual incomprehension.
The evidence that this is happening is accumulating. A 2024 Pew Research study found that Americans across political divisions were increasingly unable to identify the same top news stories from a given week, and that their news sources shared less than 5% of content. A 2023 study in the American Political Science Review found that exposure to counter-attitudinal information on social media actually increased rather than reduced partisan hostility — not because the information was wrong but because the context in which it appeared signalled tribal rather than informational intent.
What is less clear is what, if anything, can be done. The platforms that built the recommendation architectures are not going to dismantle them; the engagement advantages are too large. Public broadcasting — the institutional mechanism designed to provide common reference points in media-pluralist societies — is under funding pressure in almost every country that maintains it.
The monoculture was not good. Its content was unrepresentative, its gatekeepers were unaccountable, and its reach depended on the suppression of voices it excluded. None of this changes the structural fact that shared cultural reference is a condition of the kind of public life that democracy requires.
The question of whether cultural commons can be reconstructed without the exclusion and power asymmetry of the old monoculture is genuinely open. What is not open is whether the commons matters. We are currently running the experiment to find out what happens when it disappears.
