Sometimes Twitter can feel like Invasion of the Body Snatchers. In the middle of a bustling metropolis, you’re suddenly overtaken by the creeping realization that some of the other faces in the crowd aren’t…quite…human. They could be very close to you, making sounds that at first blush resemble human language. They could have the faces of people you know. They could even have your own face! But they’re really…bot people.
Happily, science has found a way to identify these algorithmically generated, PR-shamming, spam-faced zombies and root them out of our Twitter feeds forever.
According to a newly published scholarly study of 160,000 tweets, it’s actually fairly easy to tell the difference between a Twitter account controlled by one person, a corporate or celebrity account with multiple users, and a fully automated spambot. The difference is not in what they tweet. It’s in when they tweet.
Real people tweet in bursts that match their daily routines, which vary from person to person; most human tweets come between 7 a.m. and midnight. Corporate accounts stick to tweeting during more traditional business hours, 8 to 8, Monday through Friday. Spambots tweet randomly at all hours of the day and night. In an ironic twist, however, the researchers found that it was easier to predict the timing of the next tweet based on the previous tweet for humans than for bots. In some ways, perhaps, we’re all a little robotic.