Twitter admits its algorithms amplify right-wing politicians and news content

The study examined Twitter’s "Home" timeline.
The study examined Twitter’s "Home" timeline. Copyright AP Photo/Jeff Chiu, FILE
By Euronews
Share this articleComments
Share this articleClose Button
Copy/paste the article video embed link below:Copy to clipboardCopied

An internal study at Twitter has revealed that the platform's algorithms amplify right-wing content over left-wing content.

ADVERTISEMENT

Algorithms on Twitter are amplifying more conservative and right-wing content, an internal study has found.

In nearly all countries analysed, tweets by conservative politicians were amplified more than tweets by liberal politicians, Twitter has admitted.

Meanwhile, right-wing news outlets were also amplified for Twitter users more than left-wing news outlets.

Twitter users who choose to view content on their “home” page based on the platform’s algorithms -- as opposed to reverse-chronological order -- see more political posts, according to the study.

Twitter says they will carry out “root cause analysis” to find out why its algorithms are amplifying right-wing content.

Former United States President Donald Trump has regularly accused Twitter of silencing right-wing voices, but the internal study indicates that Republican content was promoted to users more than Democrat content.

Social media giants have previously been slammed on the left for not doing enough to combat misinformation.

The report analysed millions of tweets published by elected officials in seven countries between April and August last year.

Politicians from four European countries -- France, Germany, Spain, and the United Kingdom -- were assessed, as well as officials in Japan, Canada, and the United States.

Only Twitter users in Germany saw more content from left-wing sources than right-wing sources.

Meanwhile, right-wing politicians in the United Kingdom were significantly more likely to have their tweets exposed to more users than left-wing politicians.

But Twitter did state that amplification was not based on specific parties, more on political views and issues.

Investigators also looked at what political news content was being amplified and recommended to users by Twitter’s algorithms.

“We used public, third-party sources -- such as official institutional websites -- to identify political party affiliation,” Twitter said in a blog post.

“We did not use Tweet content to attempt to infer political views of elected officials.”

Rumman Chowdhury is Twitter’s director of machine-learning ethics, transparency and accountability.

Rumman Chowdhury, Twitter’s Director of Software Engineering, said that algorithmic amplification could have an “adverse” effect.

“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 said in the blog post.

“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.”

Twitter has said it would make its research available to third parties, but not its “raw data” over privacy concerns.

ADVERTISEMENT

“We believe it’s critical to study the effects of machine learning on the public conversation and share our findings publicly,” Twitter said.

“This effort is part of our ongoing work to look at algorithms across a range of topics,” the company added.

“We hope that by sharing this analysis today, we can help spark a productive conversation with the broader research community to examine various hypotheses for why we are generally observing comparatively more right-leaning political amplification of elected officials’ content on Twitter.”

Share this articleComments

You might also like

Donald Trump to launch social network after being banned by Twitter, Facebook and YouTube

Facebook, Twitter, and Google face questions in U.S. Congress over misinformation

Explained: Social networks face backlash over US election misinformation methods