Twitter has become one of the most popular places online for groups of people with similar interests to come together, but when Greg Marra built an automated program to manipulate that sense of community even he was surprised at how well it worked.
Marra built @Trackgirl, an experiment to see if a fake program could infiltrate and fool networks of real people, Wired reported. After a few months of running the program in 2008, Marra found that real people were sending @Trackgirl direct messages and asking how she was recovering from a fake injury the bot posted about.
Marra, who spoke about his work at the MIT-Knight Civic Media Conference last week, said the bot was built to scour Twitter for messages with running-themed keywords and then repost them as its own. It gained five followers within a few days, following them back automatically, and soon more followers started to pour in.
The ruse went a step further one day when @Trackgirl found and posted a status about hurting her ankle, Marra said.
“People were sympathising with a Python script,” Marra, a Google+ product manager, told Wired.
The results were beyond what Marra expected, the report said. While he anticipated that the bot would gain followers, he never thought people would interact with it. Other bots have been used to trick people into clicking on work-from-home scams, porn sites or illegal prescription drugs, Wired reported, but there could be greater implications for these fake Twitter posters. Roughly 35 percent of the people @Trackgirl followed ended up following her back, a “remarkable record” for a bot, the Smithsonian blog reported.
Most other bots only need users to click links and don’t need any kind of interaction, but @Trackgirl showed that these can be more than spammers. These bots could have political implications, Wired reported, and Mexico’s Institutional Revolutionary Party has already been accused of creating Twitter accounts to post identical political messages.
The practice appears to already be widespread across the social network. Wired reported that web developer Ivan Santiesteban has identified 20,000 fake Twitter accounts and a similar effort called BotBusters found a similar number.