Bot Sentinel, an analytics service, found that a small number of Twitter accounts are responsible to more than two-thirds of the hateful tweets about Meghan Markle or Prince Harry.
The pair have faced a torrent online criticism since their announcement of their departure from the royal life in January 2013. Bot Sentinel, after reviewing over 114,000 tweets, found that 70% of this comes from just 83 accounts and has been distributed in a coordinated way.
“Our research revealed that only a small number of single-purpose antiMeghan and Harry accounts were responsible for most hateful content on Twitter. However, the report says that primary accounts had access to assistance that allowed their content be repackaged and shared with accounts with a substantial following.”
“We observed that the primary accounts coordinated their activities and used various techniques to avoid detection. In other words, most of the anti-Meghan/Harry activities weren’t organic.”
The team identified 55 single-purpose accounts to be the primary hate accounts. The 28 remaining accounts were used to amplify these messages. With a total number of 187631 followers, the accounts are believed have reached 17 millions users. They also frequently linked to private blogs, Instagram accounts, YouTube channels, and YouTube channels that had similar views.
Twitter had previously suspended 40% of primary accounts. However, they managed to evade this by adding a ‘parody” to their profiles. Bot Sentinel suggests that the Twitter algorithm did not help.
“We used Twitter accounts without friends or followers during our research, and after viewing two hate accounts, Twitter’s algorithm began suggesting numerous hate accounts,” says the team.
“Twitter recommended that we follow these hate accounts on multiple occasions.”
Twitter’s algorithm has already come under fire this month. Last week, the company admitted that it was biased in amplifying tweets from politicians.
“In six out of seven countries — all but Germany — Tweets posted by accounts from the political right receive more algorithmic amplification than the political left when studied as a group,” wrote director of software engineering Rumman Chowdhury in a blog.
“Right-leaning News Outlets,” as defined by the independent organizations mentioned above, receive greater algorithmic amplification on Twitter than left-leaning media outlets.
Update:“ “Our teams are currently reviewing the accounts mentioned in this report and have taken enforcement actions, when appropriate against accounts and content that violates Twitter Rules, including potential violations by our hateful conduct and coordinated harmful activity policies,” says a Twiter spokesperson.
“Of the 50+ accounts mentioned in the report, our teams took actions on four accounts for violating our platform manipulation and spam policy. This means that many of these accounts are actually run by distinct individuals. At this time, there’s no evidence of widespread coordination, the use of multiple accounts by single people, or other platform manipulation tactics. “