Anyone who has been on Twitter for more than a few days knows that it’s rife with bots — accounts that are run by a computer, rather than a human. But while bots may be pretty easy to come by, it isn’t often that you get the chance to read through the code that makes one tick. Tonight, you’ve got your chance. The bot’s name? Jason Thorton. He’s been humming along for months now, sending out over 1250 tweets to some 174 followers. His tweets, while not particularly creative, manage to be both believable and timely. And he’s powered by a single word: Love.
Thorton is the creation of developer Ryan Merket, who built him as a side project in around three hours. Merket has just posted the code that powers him, and has also divulged how he made Thorton seem somewhat realistic: the bot looks for tweets with the word “love” in them and tweets them as its own. From Merket’s blog:
Jason tweets A LOT about the word “love” – that’s because Jason actually steals tweets from the public timeline that contain the word “love” and posts them as his own.
Jason also @ replies to people who use the word “love” in their tweets, and asks them random questions or says something arbitrary.
Merket then goes on to detail why the ease with which he could build the bot should be cause for concern:
It took me about 3 hours to code Jason, imagine what a real engineer could do with real AI algorithms? Now realize that it’s already a reality. Sites like Twitter are full of side projects, company initiatives, spam bots, and AI robots. When the free flow of information becomes open, the amount of disinformation increases. There’s a real need for someone to come in and vet the people we ‘meet’ on social sites — it will be interesting to see how this market grows in the next year.
Can social networks really vet every single user that joins? That would likely be incredibly difficult to scale, but there’s certainly room for the algorithms to improve. In any case, here are some of Jason’s most recent tweets:
And here are some of the people who fell for them: