The US election between Donald Trump and Hillary Clinton is increasingly being fought on social media platforms like Twitter, where the candidates boast millions of followers. But not all of them are human – or benign.
Automated programs are a dominant, yet often underestimated feature on the internet. Bots, as these programs are also called, make up roughly 50 percent of all online traffic. In their simplest form, bots - short for robots - fulfill benign and banal tasks, including, for instance, automatically tweeting the most current news headlines. DW and many other media organizations deploy bots doing exactly that.
However, with the prevalence of bots on the internet and the rise of social media as a tool for the dissemination of political information, it is no surprise that political actors around the world increasingly resort to these automated programs for their own, not always benign, purpose. Probably the most famous case is that of a Mexican hacker who claimed to have manipulated various elections in Latin America. Government-affiliated political bots are also said to have played a role in recent Russian elections.
Given this trend and the increasing sophistication of these automated programs, it is clear that political bots also play a role in the American presidential election - the US is, after all, the home of Twitter, Facebook and Co.
"The candidates definitely have a bunch of fake followers that can be attributed to botnets or armies of bots," said Dhiraj Murthy, a social media scholar at Goldsmiths, University of London, and author of a book on social communication in the Twitter age. And since the number of followers has become a "really important metric" for people, it is important to know how many of Trump's and Clinton's followers are genuine, he added.
Services such as TwitterAudit and BotOrNot try to approximate how many of Trump's more than nine million and Clinton's more than seven million Twitter followers are real or fake and which of those fake followers aren't simply inactive accounts, but bots.
A recent analysis by Andrew McGill in The Atlantic suggests that more 75 percent of each candidate's followers could be real people; less than four percent of the candidates' Twitter followers showed "obvious bot behavior." While that figure would be comparatively low, in absolute terms it still means that more than 200,000 Twitter followers of Trump and Clinton, respectively, are bots.
And those hundreds of thousands of bots do have a noticeable impact, Murthy explained. "If you look at any tweet Trump has posted right when he posted it and you look at the people that retweet and mention that content, a bunch of them are always bots. Because they are programmed to do that."
Amplifying and distributing messages is a crucial function in the social media ecosystem, where things can quickly be overtaken by new content and vanish. "Bots can keep certain things on the agenda and they can also propagate things quite far, depending how their networks are configured," Murthy said. "They are really important in propagation, and they have been in this election too."
But bots can also be deployed to drown out conflicting messages, noted Philip Howard, a professor at Oxford University who researches political bots. "We know that they have been used effectively to choke off debate that gets centered around particular hashtags."
He cites the beginning of the Syrian civil war as an example where the Assad regime hired a firm in Bahrain to use bots on the #Syria hashtag. As a result, said Howard, the regime flooded social media with tourism pictures of beaches, soccer scores and stories from Syrian soap operas. "For several days it was the main way that the world, including journalists, outside was getting news out of Syria."
Trump provides bot fodder
In the US election context, Donald Trump is widely viewed as the undisputed master of social media. With good reason, said Murthy, because Trump's discourse with its often inflammatory remarks provides ample fodder – not just for the media, but also for bots. "There are plenty of really mean bots on the internet that are looking for sexist, homophobic and racist language. Those bots are then finding these tweets in the context of Twitter and then following Trump and retweeting his content and mentioning him."
Of course not all political bots are malicious. There are parody bots like @DeepDrumpf or @Clinton_Bot explicitly dedicated to poking fun at the candidates. Bots like these that make transparent that there is no real person behind it are one thing. A quite different thing is when bots pretend to be human, something that is increasingly hard to ascertain.
"A very important distinction is whether people realize that there is a bot," said Cornelius Puschmann, a senior researcher at the Alexander von Humboldt Institute for Internet and Society in Berlin. "If the bot is impersonating a person and trying to pass itself off as human and that happens in a political arena, that is something generally viewed as problematic."
Non-human actors here to stay
Who exactly is behind the thousands of Twitter bots that follow Trump and Clinton is the million dollar question and difficult to trace, said Murthy. It's also not clear whether the campaigns themselves directly or indirectly deploy unattributed bots as part of their social media strategy to boost their own follower count and messaging or to attack an opponent. The Trump and Clinton campaigns did not respond to a request for comment.
What is clear, however, is that political bots are here to stay - in the US election and beyond. "The fact that non-human actors exist in social media is something that we need to get used to and something that will become more common. We all need to familiarize ourselves with this," said Puschmann.
Because of the growing importance of political bots, it's time to think about regulating them, said Howard. "I think the normal rules that govern political speech should apply, so that overtly political bots should also identify who paid for them and what their affiliations are."