Can a Twitter taunt bot defeat the trolls and save political discourse?
After every presidential debate, one thing is predictable: Twitter traffic increases, and the majority of tweets about the debate are in favour of Republican candidate Donald Trump.
A new study by Oxford University researchers this week reveals that pro-Trump tweets outnumber pro-Clinton tweets by a margin of two-to-one.
But the study also notes that a third of pro-Trump tweets aren't generated by engaged humans. They're generated by Twitter bots. By contrast, approximately a quarter of pro-Clinton tweets being bot-generated.
Twitter bots are automated social media accounts that can tweet on any topic. Some are also programmed to engage in responses with automated replies.
Fighting bots with bots
As Reed explains to Day 6 host Brent Bambury, the bots, which they call 'honeybots,' are specifically targeted toward people looking for Twitter battles.
They're arguing with robots instead of people... It's emptying the ocean with a bucket to try to waste these people's time, but it's a way of neutralizing harm.- Nora Reed
"They're named for honeypots, which is term in computing for a trap, for certain software. Mine is sort of a human honeypot," Reed says.
The honeybots send tweets without any hashtags, meaning they can only be found by someone searching out specific words. These people are commonly referred to as 'trolls.' And trolls waste a lot of time on Twitter.
Reed created the bots to waste the time of the trolls. Creating bots can take 10-15 minutes, so if desired, they take little of the inventor's time.
Setting the trap
Reed says the honeybots tweet a lot of names of politicians and columnists who will provoke a reaction.
"News outlets like Breitbart will name search themselves and then retweet anyone who's critical of them to get their fans to yell at that person for them," Reed explains.
The bots also send out tweets using words like 'feminism,' 'atheism,' and 'code of conduct,' which Reed says is popular in certain circles.
"Isn't it obvious to the people engaging with them that they're talking to a bot?" asks Bambury.
"You know, you'd think it would be," says Reed. But people tend to simply react to the tweets without looking at the metadata or Googling the user.
Often, Reed says, they're responding to a retweet from someone they follow.
"Occasionally they figure out that it's automated, because, again, this has less replies in it than a Magic 8 Ball," says Reed.
Trolling the trolls
"What I think it sheds light on is the fact that none of these people are actually interested in having a conversation," says Reed.
Reed goes on to explain that some trolls just want to subject people to their talking points. Some will copy and paste a list of comments from a notepad document, while others just re-paste from a list of image files that they use over and over again.
Reed says creating the bots was like creating a "sort of control subject."
Reed says that people often dismiss complaints of Twitter harassment of real people, saying that the person may have done something to provoke the abuse. But Reed says her bots show the reality of online harassment by demonstrating that the same abuse is hurled at automated Twitter accounts.
Reed also hopes that the bots can create a sort of catharsis for people who have gone through that type of harassment.
"It also wastes people's time because they're arguing with robots instead of people... It's emptying the ocean with a bucket to try to waste these people's time," says Reed, "but it's a way of neutralizing harm."