The experiment involved over 1,000 users, but platforms operate at vastly larger scales serving billions globally. Extrapolating experimental findings to understand real-world impacts requires careful thought about how effects might accumulate and interact when algorithmic manipulation affects entire populations rather than small research samples.
If one week of subtly altered feeds produced polarization equivalent to three years of natural change among 1,000 users, what happens when similar algorithms affect billions of users over months and years? The cumulative effects could be staggering—potentially accelerating societal divisions at rates far exceeding anything in human history.
Individual effects measured in the study proved substantial but might seem manageable in isolation. Shifts of two degrees on a 100-point feeling thermometer don’t make someone violent or irrational. But when billions of people simultaneously experience such shifts, the aggregate societal impact includes measurable increases in political violence, democratic backsliding, and governance dysfunction.
Network effects compound individual changes. As more people become polarized, they influence others through social interactions both online and offline. Polarization becomes self-reinforcing as societies sort themselves into hostile camps that interact primarily within rather than across political divides. Algorithmic acceleration of this process could push societies past tipping points where divisions become effectively irreversible.
The billion-user scale also means that even small percentage effects produce massive absolute numbers. If algorithmic polarization makes just one percent of users measurably more supportive of political violence, that still represents tens of millions of people globally whose attitudes toward democratic norms have been harmfully influenced by corporate algorithms optimized for engagement rather than social cohesion.