Bail et al. (2023) — Exposure to opposing views on social media can increase political polarization
Bail, C.A. et al. (2023). Exposure to opposing views on social media can increase political polarization. Proceedings of the National Academy of Sciences. https://www.pnas.org/doi/10.1073/pnas.2213020120
What this source contributes
A field experiment testing whether algorithmic exposure to opposing political views reduces polarization. The finding runs against the intuition behind most proposed remedies: exposure to opposing views through social media recommendations does not reduce polarisation, it increases it.
The mechanism is that the emotional response to disagreement is itself an engagement signal, which the algorithm reads as a reason to surface more of the same. What looks like a corrective becomes another input to the optimisation.
Analytical function in Recommender Systems
Supports the entry’s central claim that optimising for engagement produces polarisation rather than correction. Important because it forecloses the obvious fix: the problem is not that people see too little disagreement.
Related entries
- Recommender Systems — exposure to opposing views as an engagement signal rather than a corrective