Researchers outline the contours of political polarization on social media

If you thought your Twitter timeline is a mess, imagine what the social media map of the U.S. political landscape looks like.

Researchers at Graphika, a network analysis firm in New York, mapped out the political landscape in U.S. on Twitter as part of a MIT Technology Review study.

John Kelly is the CEO and Camille Francois is research director of Graphika. The firm specializes in detecting and analyzing social media threats and media manipulation.

But Kelly and Francois found that the loud and consistently active political "far right" and "far left" are larger influencers. They also found the political middle to be quite weak in comparison, making it easier for "foreign actors" to weaponize the timeline, a finding researchers call "striking."

"What we see now very clearly, is that this possibly natural polarization or our tendency toward polarization has been effectively weaponized by actors that have become extremely good at both the automation of messaging," Kelly told Spark host Nora Young.

The visual representation is based on when influential political accounts tweet about politics and how often, the ways they amplify similar views, and what other accounts they follow.

The graphic "is a network map of the follower and the community assembled around traditionally elected American politicians," said Francois.

"What we use as seeds to create this landscape is a list of every elected American congresswoman and congressman. It's sort of a reflection of the communities gathered around traditional American politics."

The map is colour-coded based on views—left, centre and right—and made up of bubbles that represent individual Twitter accounts.

The larger the bubble, the more followers the account has—such as news media, celebrities, politicians, and Twitter personalities.   

"It's kind of like the normal volume of conversation happens in the middle, [and] both extremes are, a) further away from that middle and a little bit disconnected from it, and b) so much louder than that middle," said Francois.

Kelly and Francois also studied how automated phony accounts disrupt the online political landscape on Twitter.  

"You've got this automated amplification using bots and programmed messaging that's really designed to appeal to the political extremes and that's not trying to do much in the middle at all," Kelly said.

The researchers looked to the Trump-promoting Twitter star, Jenna Abrams, as an example. The popular Twitter persona turned out to be a fake product of the Russian intelligence firm, the Internet Research Agency.

The Abrams Twitter account had more than 70,000 followers before it was outed as a propaganda tool. The tweets were conservative, often racist and segregationist, and featured light commentary about popular culture. The Abrams account was often quoted in mainstream media stories.