Twitter Reports Algorithmic Bias Toward Right-Leaning Content

Twitter has learned its algorithm amplifies right-leaning political content, the social media company revealed in a recent blog. In a study conducted this year, Twitter sought to identify how machine learning (ML) may impact the promotion of political content on the site.

The algorithm in question pertains to Twitter’s Home timeline, which displays Tweets from the user’s followed accounts as well as recommended content. Content is recommended based on accounts and tweets with which a user interacts.

Twitter’s study observed content from April 1^st and August 15^th, 2020, largely from elected officials in seven countries (Canada, France, Germany, Japan, Spain, the United Kingdom, and the United States). The site also studied the effect of its recommended algorithms on the amplification of political content from news outlets.

Upon analysing their findings, Twitter published that they had found a “statistically significant difference favouring the political right wing” among all countries observed except Germany. Right-leaning news outlets saw “greater algorithmic amplification” on Twitter compared to left-leaning counterparts.

The greatest discrepancy between right- and left-leaning algorithmic amplification was seen in Canada, where Conservatives saw a 167 per cent boost compared to the left-leaning Liberals’ 43 per cent.

Although Twitter is now aware of the bias, the cause is yet to be determined. In a post explaining the findings, Rumman Chowdhury and Luca Belli explained that algorithmic amplification is not inherently problematic.

“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 post explained.

Twitter also established plans to make their data available upon request to “third party researchers who wish to reproduce [Twitter’s] main findings and validate [their] methodology”.

According to their blog, Twitter hopes that releasing the information will “spark productive conversation” and incite further research into the cause of such algorithmic biases.