Smithsonian.com

A Twitter bot pretended to be a runner. Photo: Flickr user hans s

Greg Marra’s Twitter bot @Trackgirl found and “repurposed” others’ tweets about running, followed five people a day, and followed anyone who followed her. She had a remarkable record for a bot: 35 percent of the people she followed followed her back. One day, she stole a tweet about being injured, and her creator found out how real her followers thought she was, Wired reports:

Soon after, her followers wanted to know if @trackgirl was on the mend. “People were sympathizing with a Python script,” says Marra, a Google+ product manager.

Turns out people aren’t so good at ferreting out real human beings. “Social bots” like @Trackgirl don’t need to get you to click on links, they need you to believe they’re real, so that when they post, say, a message about supporting a particular political candidate, their audience takes that opinion seriously:

“Social bot attacks are actually about building a trust relationship,” [Marra] says. “So they’re very slow attacks. Your goal is not on the first day that you launch your bot army to do anything. Your goal is over months to build up history, build up credibility and build up an audience.”

Sneaky. Also, good advice for anyone trying to amp up their Twitter follower count.

More from Smithsonian.com

The Dinosaurs of Twitter