Twitter’s algorithm boosts rightwing political content

The study analysed whether Twitter’s recommendation algorithms boost political content and focused on tweets from elected officials and political content from news outlets.

Twitter analysed millions of tweets between 1^st April and 15^th August 2020, from accounts operated by elected officials in Canada, France, Germany, Japan, Spain, the United Kingdom, and the United States.

The company used the data to test whether or not those tweets were amplified more on its algorithmically ranked home timeline than the reverse chronological feed and whether there was variance within a political party.

It also analysed hundreds of millions of tweets containing links to articles shared by people on Twitter during the same time period, to study the algorithmic amplification of news outlets. Researchers excluded tweets linking to non-political content, and media outlets were categorised based on media bias ratings from independent organisations AllSides and Ad Fontes Media.

The research found that in six of the seven countries studied – all but Germany – tweets posted by accounts from the political right received more algorithmic amplification than the political left when studied as a group.

Twitter’s algorithms also amplified right-leaning news outlets more, compared with left-leaning media, according to the study.

Twitter’s algorithmic timeline was found to amplify political content from elected officials more than political content on the company’s reverse chronological timeline, regardless of party or whether the party is in power.

Establishing why certain political content is amplified is a “difficult question to answer as it is a product of the interactions between people and the platform”, Twitter’s director of software engineering Rumman Chowdhury and staff machine learning researcher Luca Belli wrote in a blog posted on the company website.

Chowdhury and Belli said: “This research study highlights the complex interplay between an algorithmic system and people using the platform. 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.

“Further root cause analysis is required in order to determine what, if any, changes are required to reduce adverse impacts by our home timeline algorithm.” 

The research was conducted by Ferenc Huszár (Twitter and the University of Cambridge), Sofia Ira Ktena (DeepMind Technologies), Conor O’Brien (Twitter), Luca Belli (Twitter), Andrew Schlaikjer (Twitter), and UC Berkeley’s Moritz Hardt, who carried out the work while consulting for Twitter.