Twitter Study Finds Algorithm Boosts Right-Leaning Content Over Left
Twitter's internal research shows its algorithm amplifies right-wing political content more than left-wing content in six out of seven countries studied. The company admits it does not know why this happens and is seeking outside researchers to help find the root cause.
Twitter reveals that its algorithm actively amplifies right-leaning political content over left-leaning content in six out of seven countries studied. Researchers analyze millions of tweets from elected officials between April and August 2020 to see how political content performs on the algorithmically ranked Home timeline compared to a reverse chronological feed. The only country where this right-leaning preference does not appear is Germany.
This finding contrasts with common complaints from conservative figures who frequently accuse the platform of anti-conservative bias. However, Twitter insists that algorithmic amplification is not inherently problematic unless it stems from the actual construction of the algorithm rather than natural user interactions. The company describes Twitter as a sociotechnical system where algorithms respond dynamically to real-world events and user behavior.
Despite conducting this extensive internal study, Twitter states it does not fully understand what causes this political imbalance. The company plans to conduct a root cause analysis to determine whether the issue is an unintended model bias, a reflection of how people tweet, or a combination of both factors. To help solve this mystery, Twitter now seeks to collaborate with external researchers to dig deeper into the algorithm's unexpected behavior.