With over 330 million monthly active users, Twitter has become an increasingly important source of news and information. Nevertheless, the platform's algorithm has repeatedly come under fire for selectively determining which tweets users see. While some critics argue that the algorithm promotes negative content and amplifies extremist views, others suggest that Twitter is trying to provide users with relevant content.
Twitter’s Algorithm Accused of Bias
In 2021, Twitter published a study showing that, in six out of seven countries, tweets from right-wing political accounts received greater algorithmic amplification over those from left-leaning accounts. Moreover, right-wing content published by news agencies had a higher chance of appearing and receiving engagement due to its suitability for Twitter's algorithm. Twitter stated that the reasons for this bias are difficult to pinpoint, as they are contingent upon interactions between users and the platform.
Research published in the journal PNAS supports the idea that negative and emotional language has a higher chance of going viral on social media platforms, including Twitter. This raises serious concerns: Could Twitter's algorithm deliberately promote negative content and amplify the reach of extremist viewpoints?
Exploiting Twitter’ Algorithm to Shape the Public Opinion
Undoubtedly, some individuals or groups can exploit Twitter’s algorithm to increase interaction and raise the odds of certain tweets and accounts appearing on users' home pages. They can recruit “electronic flies” to create fake interaction with specific accounts, which can fool the algorithm into considering those accounts to be important. By adopting a publishing method that complies with the algorithm's standards, they can deceive it and drive it to amplify the reach of some tweets, thus influencing and even shaping the public opinion.
It is not certain whether the algorithm is the reason behind the increase in the reach of some suspicious accounts, or not. The exploitation of the algorithm, on the other hand, is very likely to occur.
How does Twitter’s promote Tweets?
Twitter’s algorithm controls what users see and how they see it. It prioritizes tweets based on several criteria, including geographic location, reach, personal interests, trending topics, and whether the post contains media or not.
Since 2019, Twitter has enabled users to follow specific topics in addition to accounts. Twitter then recommends tweets based on the topics users have shown interest in, using machine learning technology to sort content based on different arrangements.
According to Hootsuite, posting tweets at least once to twice a day and a maximum of three to five times a day can increase the chances of their appearance on users' main pages.
The Return of the "For You" Page on Twitter
One of the new updates on Twitter is the return of the "For You" page, which features tweets, accounts, suggested topics, and other content that the Twitter algorithm believes a user might be interested in. The more people use Twitter, the more accurate its filters.
Starting from April 15, only tweets from accounts that paid for verification will appear on the "For You" page. This raises concerns about accounts that spread misinformation and unverified information on Twitter, as they can pay for verification and obtain the preference for their publications to appear to users, contributing to the spread of misinformation faster and on a larger scale.
The Twitter algorithm filters tweets based on several criteria, including:
- Trend: Priority is given to topics that are popular now, rather than topics that have been popular for a while.
- Relevance: Twitter builds a user profile based on past actions, such as the account's tweets. It then tracks the number of tweets the account has posted in relation to a topic. Twitter gives priority to the accounts that the user interacts with most often and the topics they follow and interact with the most.
- Location: Location is a key criterion upon which Twitter filters tweets for users.
- Engagement: Twitter prioritizes a tweet based on its popularity and how people in a user's network interact with it. If a user follows a particular topic, Twitter will look at the number of people who tweet, retweet, reply, and like tweets about that topic.
- Mixed Media: Twitter categorized tweets based on the type of media they contain, such as images, videos, GIFs, or polls.
Tweets by Sam Youssef and Edy Cohen
Misbar’s investigation revealed that Twitter has been giving tweets by Edy Cohen and Sam Youssef and advantage even if you do not follow their accounts.
The report by Misbar delves into the methodology used by the two accounts to publish their tweets and its impact on the content promoted to Twitter users. The report also considers the possibility of third-party entities systematically creating fake interactions with the accounts of Cohen and Youssef to deceive the Twitter algorithm, increase tweet reach, and manipulate public opinion.
The analysis focused on the criteria for promoting tweets in the Twitter algorithm and examined the data of 2,308 tweets published by Sam Youssef's account and 728 tweets published by Edy Cohen's account between March 1, 2023, and April 11, 2023.
It must be said that the two accounts have a reputation for publishing controversial and suspicious content, often linked to Israeli propaganda and conspiracy theories.
Examples of Sam Youssef's Tweets
Sam Youssef's account was among the proponents of the conspiracy theory that suggested the possibility of using HAARP technology weapon to cause natural disasters after the earthquake in Turkey and Syria in February. Misbar published a report debunking the theory at the time, stating that the HAARP project could not cause natural disasters or manipulate human thoughts.
Examples of Edy Cohen's Tweets
Edy Cohen's account has promoted Israeli occupation propaganda and false narratives in the past, with some of his content debunked in a previous article by Misbar.
An Analysis of Tweets by Sam Youssef and Edy Cohen
The two accounts follow trends and tweet about the latest events, whether by publishing news or expressing their opinions. Consistent posting, at least once a day, increases the likelihood of tweets reaching other users' homepage. When Misbar analyzed the data of the two accounts, we found that they continued to tweet throughout the study period, posting at high rates without refraining from tweeting on any day.
According to the graphs, Sam Youssef tweets more than Edy Cohen in a single day. For instance, on March 10, 2023, Sam Youssef tweeted 148 times, while Cohen tweeted only 15 times, with replies to tweets included in their total numbers as shown in the figure below.
The hashtags used by the two accounts give an idea of the topics they are interested in publishing, which often relate to issues that are polarizing the Arab public opinion.
Both accounts use hashtags related to different topics. For example, Sam Youssef used the hashtag "urgent" 113 times in 42 days, while Edy Cohen used the hashtag "Kingdom of Saudi Arabia" 11 times in the same period. Meanwhile, the second most used hashtag was "#Palestine_Not_My Cause."
When comparing the topics discussed by the two accounts, Misbar found that their tweets during the 42-day period focused on issues related to Israel, Iran, Saudi Arabia, and Syria.
Sam Youssef used the hashtag "Israel" 49 times, while Edy Cohen used it only five times.
Sam Youssef used the hashtag "Iran" 20 times, while Cohen used it once.
Sam Youssef used the Saudi hashtag 31 times, while Edy Cohen used it four times.
Sam Youssef used the Syria hashtag 26 times, while Cohen used it four times.
The two accounts diversify their use of media. For example, Sam Youssef used about 1,912 tweets with a picture and 2,300 links within 42 days.
Meanwhile Edy Cohen used 141 photos, two videos, and 579 links.
Through the reliance on Python language and the use of NLTK tools that analyze data based on natural language processing, the tweets studied in this investigation by Misbar were classified in terms of emotional language into categories: positive, negative, and neutral.
After calculating the rate of likes and retweets in each of these categories and analyzing the data again in Python, we found that the content that used negative emotional language in tweets by the two accounts had the highest rate of retweets and likes.
Edy Cohen’s Account
Sam Youssef’s Account
Based on Misbar's findings, it appears that the Twitter algorithm is built in a way that can be exploited. Twitter can thus become a tool for promoting suspicious content and influencing public opinion by targeting it with specific propaganda, as is the case with the accounts of Edy Cohen and Sam Youssef.
However, it cannot be excluded that certain individuals and groups can create a systematic fake interaction with the accounts of Cohen and Youssef in order to deceive the Twitter algorithm and amplify the reach of their tweets to the target group, with the aim of misleading and influencing public opinion.
Misbar’s Sources:
