Changing Facebook and Instagram feeds from algorithmic to chronological order changed activity and exposure but did not significantly change measured polarization or other key attitudes over three months. This cuts against a simple claim that the experiment showed algorithms increasing polarization.
Although the chronological feed altered exposure to political content, it did not significantly alter levels of issue polarization, affective polarization, political knowledge, or other key attitudes during the 3-month study period.
Abstract
Changing Facebook and Instagram feeds from algorithmic to chronological order changed activity and exposure but did not significantly change measured polarization or other key attitudes over three months. This cuts against a simple claim that the experiment showed algorithms increasing polarization. Bearing on the claim: contradicts. Verification note: A 2024 erratum concerns some Instagram untrustworthy-source measurements; it states treatment-effect estimates were not affected.