
The bots on social media are talking. And they’re talking about one guy in particular.
You’ve probably heard a lot about bots – automated social media accounts that have been used to influence human opinions on politics and other topics. You might have heard that the Russians used them to influence the 2016 U.S. presidential elections, often by spreading misinformation and amplifying ideas thought to support a certain political agenda.
As pollsters, Ipsos often asks humans what issues are top-of-mind or worrisome. But if the bots are driving some of our social media conversations, what would happen if you polled the bots? That’s one side project of Ipsos’ new polling site, Political Atlas, developed in partnership with the U. In addition to traditional polling data and the “Crystal Ball” predictions from famed polling analyst Larry Sabato, the site tracks social media conversation. It drills down into what, specifically, the bots are talking about and which races are they most trying to influence. It’s worth noting that the site doesn’t try or claim to identify the source of the bots. Nor do we suss out why they are trying to influence the race – that’s more of a law-enforcement issue.
Trump tops conversation topics
Nationally, President Donald J. Trump is the leading topic of social conversation for both humans and bots, which are driving 12 percent of the conversation. The economy, the Supreme Court, healthcare and foreign policy round out the top five topics in sheer number of social posts. Trump is the lead topic in nearly every state, as well. The exceptions are Colorado and Vermont (healthcare) and Nebraska (economy).
To determine these leading topics, Ipsos analyzes millions of tweets during the last 60-day period using artificial intelligence and language processing algorithms to understand the topics Twittered, Facebooked and Instagrammed about. To separate the bots from the men and women, Ipsos uses a series of artificial intelligence algorithms to identify suspected bot accounts based on both stated and inferred location, frequency and topics of posts, social network analysis and more.
The result is a first-ever, real-time Bot Poll.
Is the social media data predictive? Will overall social popularity (from bots and actual humans) lead to campaign victories? Not necessarily, but it tells an important part of the larger story. “As political forecasting evolves, it’s important to draw insights from as many data sources as possible,” says Cliff Young, President, US, Ipsos Public Affairs. “Polling will always be an important part of the equation, but we now have access to social media feeds, location-based intelligence, search volume data and more.”
Why bots are worth polling
The bots are an especially interesting phenomenon to study. Bots can provide a critical accelerant for both misinformation and true stories, according to an MIT study that “found that bots accelerated the spread of true stories at approximately the same rate as they accelerated the spread of false stories.”
Whether the stories are true or not is often beside the point. “One common denominator is that they amplify issues or highlight topics that are divisive in nature,” said Craig Silverman, media editor for Buzzfeed and noted expert on misinformation. The bots, he says, “are trying to push people into camps or make ideas seem more popular than they are.”
This artificial influence can play a role in close elections, whether its nefarious actions from foreign governments, or more benign applications like campaigns mechanizing these platforms to get their campaign messages out and raise cash for their war chests. The bots are also active on non-election topics, spreading misinformation about public health and other campaign platform topics, for example.
As the election approaches – and as we look toward 2020 – keep an eye on Political Atlas, which will be keeping an eye on the bots and helping shed some light on their activity. Silverman says that can certainly help, and that increased awareness and growing efforts on the part of tech companies, academics and government agencies to combat this is a good sign. But he warns that people on platforms like Twitter now see “a bot around every corner” and are accusing human trolls of being bots. That paranoia, he says, is a “victory for the people trying to sow distrust.”