Analysis of a study conducted by the Pew Research Center that was published on Monday showed that a majority of links shared on Twitter are shared by bots, rather than human users.
Focusing on popular websites and media outlets, ‘as measured by the number of links posted on Twitter to their content,’ the study showed that 66 percent of links shared on Twitter to those sites came from what seemed to be automated accounts, as determined by an app called Botometer.
However, it’s important to note that the study did not address who was responsible for the creation of the bots found to be disseminating the most shared stories, or the reach of the links found to be tweeted by bots.
This means the study in no way reflects how many human users saw, clicked through or engaged with tweets generated by bots that shared content.
A Pew Research Center study showed that 66 percent of links shared on Twitter to ‘popular’ sites came from automated accounts, as determined by an app called Botometer
The study broke down the ‘popular’ websites it analyzed into six categories, comprised of adult content, sports, celebrity, commercial products or services, organizations or groups, and news and current events.
Overall, it found that 66 percent of tweeted links to the most linked sites in each category were shared by automated accounts, rather than human users.
The percentages varied drastically from within each category, with automated accounts being found to tweet 90 percent of links to the most tweeted adult content, and 53 percent of links to organizations or groups content being shared by bots.
The categories of sports came in second-highest for bot-tweeted links at 76 percent, followed by commercial products or services, at 73 percent.
Bot-tweeted links for the most ‘popular’ news and currents sites echoed the average, overall, at 66 percent.
The category of celebrities was found to have 62 percent of tweeted links to content coming from automated accounts.
These findings are interesting, in part, because total number of shares is one metric used by sites to measure audience engagement with their content.
It begs the question of who is behind the creation of these bots that are sharing content from these recognized ‘popular’ websites.
One possibility is that the sites, themselves, may have created automated Twitter accounts to share their links on the platform, thus artificially propping up apparent engagement.
It also leaves the question open of how to determine whether a Twitter account is automated, or run by a human.
Overall, the study found that 66 percent of tweeted links to the most linked sites in each category were shared by automated accounts, rather than human users
This is something that researches at the University of Southern California and Indiana University are grappling with.
Academics at the two institutions have both contributed to the creation of an app which analyzes Twitter accounts and gives them a resulting bot score.
The app, called Botometer, is in its second iteration, but is not perfect.
The website for the current version of the application, identifying itself as a joint project of the Indiana University Network Science Institute (IUNI) and the Center for Complex Networks and Systems Research (CNetS), includes a disclaimer that ‘Botometer often categorizes “organizational accounts”, like @BarackObama, as bot accounts.’
The Pew Research Center utilized Botometer to determine whether a link was shared by a bot or a human in its analysis, but did not disclose what percentage likelihood that an account was a bot was required for the Center to classify an account as an automated account in its study.
For context, this reporter ran the Botometer analysis on her own Twitter account, and the results came back that the account was 47 percent likely to be run by a bot, which it is not.
The Pew Research Center utilized Botometer to determine whether a link was shared by a bot; For context, this reporter ran the Botometer analysis on her own Twitter account, and the results came back that the account was 47 percent likely to be run by a bot, which it is not
The study by the Center analyzed 1,220,015 tweeted links to 2,315 ‘popular’ websites, collected between July 27, 2017 and September 11, 2017.
To determined what constituted a ‘popular’ website for inclusion in its analysis, ‘the Center identified nearly 3,000 of the most-shared websites during the first 18 days of the study period and coded them based on a variety of characteristics.’
‘After removing links that were dead, duplicated or directed to sites without sufficient information to classify their content, researchers arrived at a list of 2,315 websites,’ the Center said.
The research group then broke those sites down into the six categories described above.