Right-wing politicians' tweets are more likely to be amplified than left-wing politicians' tweets on Twitter
On October 21, 2021, Twitter announced the results of a study that analyzed tweets from members of parliament from seven countries around the world, including Japan. This revealed that in six countries other than Germany, 'the tweets of right-wing lawmakers are more susceptible to algorithmic amplification than the tweets of left-wing lawmakers.' Twitter says it's unclear whether this bias is due to the user or the bias of the algorithm itself.
Algorithmic Amplification of Politics on Twitter
(PDF file)
Examining algorithmic amplification of political content on Twitter
https://blog.twitter.com/en_us/topics/company/2021/rml-politicalcontent
ISPs have so much data on you --Protocol — The people, power and politics of tech
https://www.protocol.com/policy/isp-ftc-data
Since 2016, Twitter has been able to choose whether to order the tweets displayed on the timeline in the order of new arrival or to rank them by an algorithm. Twitter's algorithm for selecting tweets to be displayed on the user's timeline is designed to select content that each user is likely to be interested in from the accounts that the user followed and the accounts that they frequently interact with. ..
To find out how this algorithm affects political activity on Twitter, Twitter uses a timeline in order of new arrival and a timeline selected by the algorithm to display content sent by politicians. We conducted a survey to verify whether it changes to.
The survey period is from April 1st to August 15th, 2020, and the analysis targets political tweets posted by personal accounts of politicians in the United States, Japan, the United Kingdom, Spain, Germany, France, and Canada. The politicians were elected members, such as members of the House of Representatives in Japan and members of the House of
The study found that 'algorithms tend to amplify political tweets by elected politicians,' comparing the timeline chosen by the algorithm with the timeline in new order. The amplification effect was irrelevant to whether the politician was a ruling party member in power or to the party to which the politician belonged.
In addition, in six countries except Germany, tweets posted by right-wing politician accounts were found to be more algorithmically amplified than left-wing politician tweets.
Twitter emphasizes that the results of this survey only show a bias in the amplification results, not that the algorithm was the cause of the bias. 'It's unclear what causes certain content to be amplified,' said Rumman Chowdhury, director of software engineering at Twitter. I think there is, but I don't know exactly what it is. Anyway, I thought it was important to share this information. '
According to Chowdhury, the success of conservative content is due to the fact that some right-wing politicians are good at marketing their idealism, and right-wing, assuming that algorithmic bias is not the cause. It seems that he has succeeded in gaining support for his idealism.
On Twitter, there was a fierce debate about how to publish the results of this research. This is because such news articles are easily misunderstood, and there is concern that by the time the research results reach a large number of people, the content will be considerably diminished. Nonetheless, Chowdhury said he thought that publishing these findings along with the data would help researchers around the world study and verify political bias on Twitter. Said.
Related Posts:
in Web Service, Posted by log1l_ks