The Algorithm Decides What Democracy Looks Like
On any given day, the average person in a developed democracy consumes more political information than a citizen of the 18th century would encounter in a lifetime. The difference is not just quantity — it is curation. Every headline, every video, every post that reaches a user's screen has been ranked, filtered, and amplified by algorithms designed not to inform, but to maximise engagement. The result is a political information environment that is systematically distorted toward the outrage-inducing, the emotionally charged, and the divisively partisan.
The Pew Research Center found that over 50% of Americans regularly get news from social media. The Facebook Widely Viewed Content reports have repeatedly shown that misinformation outperforms accurate news in reach and engagement. These are not accidents of design — they are features.
How Algorithms Shape Political Reality
The mechanism is straightforward, even if its consequences are profound. Social media platforms — Facebook, YouTube, X (formerly Twitter), and TikTok — use recommendation algorithms that optimise for the metric they can measure most reliably: time spent on platform. Research consistently shows that emotionally provocative content, particularly content that triggers anger or fear, drives more engagement than neutral or positive content. The algorithms learn this and amplify accordingly.
The result is what researchers call the outrage spiral: politically charged content travels further, faster, and reaches larger audiences than nuanced analysis. Fringe views that would once have struggled to find an audience are now algorithmically boosted to millions. Political moderates — the majority in most democracies — are systematically underrepresented in the content that reaches their peers.
The Evidence on Political Polarisation
A landmark 2023 study published in Science directly examined Facebook's algorithm's effects on political polarisation during the 2020 US election. Researchers collaborated with Meta to run controlled experiments on millions of users and found that algorithmic ranking significantly increased exposure to cross-cutting political content — but also to highly partisan content, suggesting the relationship between algorithms and polarisation is complex. The study's conclusion: the algorithm is neither the only cause of polarisation nor an innocent bystander.
What is clearer is the speed at which misinformation spreads. A seminal MIT study on Twitter found that false news spread six times faster than accurate news, reached more people, and penetrated deeper into social networks. The mechanism is not bots — it is human beings, rewarding emotional content with shares and retweets regardless of its accuracy.
Democracy Under Pressure
The political consequences are real and documented. The Freedom House Freedom in the World report has documented a 17th consecutive year of global democratic decline. Researchers at V-Dem Institute have linked the rise of social media to the erosion of democratic norms in multiple countries, citing the use of platforms to spread disinformation, coordinate election interference, and suppress political opposition.
The 2016 US election, the Brexit referendum, the 2018 Brazilian election, and elections in the Philippines, Hungary, and India have all featured social media disinformation campaigns that reached significant portions of the electorate. In each case, the platforms provided the distribution infrastructure that made mass manipulation possible at a fraction of the cost of traditional media campaigns.
What Can Be Done
The policy conversation has matured considerably. The EU's Digital Services Act now requires large platforms to conduct risk assessments for their algorithmic systems and provide researchers with data access — a significant step toward accountability. The UK's Online Safety Act introduces new duties around content moderation. In the US, proposed reforms to Section 230 would remove legal immunity for algorithmically amplified content.
Academic researchers and civil society groups like the European Digital Media Observatory and First Draft are developing media literacy programmes and real-time fact-checking infrastructure. The argument is not that social media is inherently anti-democratic — it can also mobilise voters, amplify grassroots movements, and hold power to account. The argument is that unregulated algorithmic amplification of engagement-maximising content is incompatible with a healthy public sphere.
The Stakes
Democracy is, at its core, a system that requires citizens to share enough of a common reality to make collective decisions. When the information environment fractures into algorithmically curated filter bubbles, that shared reality erodes. Societies cannot deliberate about shared problems when they disagree on the basic facts of those problems.
This is not an abstract concern. It is the central political challenge of the 2020s. The question is not whether social media shapes democracy — it demonstrably does. The question is whether democratic societies will choose to shape social media in return. For ongoing coverage, follow the Knight Foundation's democracy research and the work of the Shorenstein Center at Harvard.