Twitter has released a study that analyses if its recommendation algorithms amplify political content on users’ timeline and the reasons why it does that. Twitter’s ML Ethics, Transparency, and Accountability (META) research team analysed the tweets posted by elected officials between April 1 and August 15, 2020. These officials belonged to Canada, France, Germany, Japan, Spain, the United Kingdom, and the United States. They also studied tweets containing links to news articles shared by people on Twitter. The analysts found that tweets posted by the political right were amplified more than those listed by the political left in all countries except Germany.

Similar was the case when it came to shared links of news outlets. Twitter analysed the links to content from news outlets and not the tweets themselves. The researchers found that right-leaning news outlets received more algorithmic amplification than left-leaning news outlets. Twitter used a classification from third-party researchers to classify news as left-leaning or right-leaning.

Twitter explained that users have been able to choose between viewing algorithmically ordered tweets first in the home timeline or viewing the most recent tweets in reverse chronological order. Twitter noted that an individual sees tweets on their home timeline based on how they interact with the algorithmic system as well as how the system is designed.

The study also found that Tweets about political content from elected officials, regardless of party or whether the party is in power, see algorithmic amplification when compared to political content on the reverse-chronological timeline. Further, two people from the same party would not necessarily see the same amplification as Twitter does not see affiliation or ideology while recommending content.

Rumman Chowdhury, Director of Software Engineering at Twitter in a blog post noted, “Algorithmic amplification is not problematic by default all algorithms amplify. Algorithmic amplification is problematic if there is preferential treatment as a function of how the algorithm is constructed versus the interactions people have with it.” She noted that further root cause analysis is “to determine what, if any, changes are required to reduce adverse impacts by our Home timeline algorithm.”