August 16, 2017


Down the Breitbart Hole (WIL S. HYLTONAUG. 16, 2017, NY Times Magazine)

Amid all the speculation about Breitbart, one of the more grounded analy­ses is a project at Harvard, where a law professor named Yochai Benkler has been studying the company's rise and influence. A few weeks ago, I met up with Benkler in his office, and we sat at a long table cluttered with books and papers to discuss his findings. Benkler was a muscular figure in his early 50s, with cropped gray hair and a trim gray beard, and he wanted to make clear from the outset that his work on Breitbart was A. the collaborative undertaking of more than a dozen interdisciplinary colleagues throughout Harvard and M.I.T., and B. a total mistake.

''We didn't set out to study Breitbart,'' he said. ''Breitbart came from the data, not the other way around.''

The project, Benkler said, began with a friendly argument about how the internet works -- whether it serves mainly as a distribution network for the articles on major media, or if small blogs and websites can funnel their own stories back into the mainstream press. ''I had taken the position that it created significant democratization and open pathways for diverse voices to speak,'' Benkler said. His friend and colleague at M.I.T. Ethan Zuckerman disagreed. ''Ethan thought you saw more amplification and circulation of stories from the mainstream.''

Normal sorts of people might settle this debate by shrugging that nobody can say for sure and ordering another round. Benkler and Zuckerman decided to build a colossal database called Media Cloud and spend the next decade hoovering up websites to see how information travels. Benkler brought in a programmer named Hal Roberts, and they began to examine the coverage of specific stories. In 2013, when George Zimmerman was acquitted in the death of Trayvon Martin, Zuckerman and a team at M.I.T. examined how various media outlets influenced public opinion. The data showed that for all the attention given to local news and activism, the mainstream media continued to shape major turning points in the story. In the debate between Benkler and Zuckerman, this was: point, Zuckerman.

But last spring, as Donald Trump was hurtling toward the Republican nomination in defiance of every analyst everywhere, Benkler got a call from the Open Society Foundations. This being an article about right-wing media, let's insert the disclaimer that O.S.F. was created by George Soros, the billionaire activist who haunts the fevered imagination of the right in about the same way the Koch brothers terrify the left. At the time, nearly everyone assumed that Clinton would win the election, but people at O.S.F. were still intrigued by the rise of Trump. ''They said: 'Something different is happening,' '' Benkler recalled. '' 'We need to understand what.' ''

Benkler made some tweaks and pointed his machine at the election. The list of websites it scooped up each day was dizzying. ''We begin with several thousand sources,'' Benkler said, ''and then we look at links in all those stories, and we crawl out to grab them. If they fit the keywords that we're looking for -- any candidate name, anything around the election -- we put them in the data set. Then we go to the links in those stories, and draw in those, and go to their links. Usually by the time we've done this 15 times, there are no new stories.''

At the same time, Benkler's team developed a method to determine the political association of any website's audience. If an article was posted by people who retweeted Hillary Clinton, they characterized the publication as one with a liberal readership. If an article was posted by Trump retweeters, they assumed the publication's readership was conservative. I can hear you thinking, ''Retweets do not equal endorsements,'' but it turns out they mostly do. But when they encountered articles linked to by supporters of both candidates, they looked to see which side linked more often, then characterized the audience as center-left or center-right.

Benkler fiddled with a laptop to show me how this looked in practice. He pulled up an image of a messy blue and red scribble against a white background. On closer inspection, this turned out to consist of thousands of tiny dots. Each represented at least one article in their database and was shaded red or blue to indicate the political association of its readers. The more times a website's article was shared on Facebook, the bigger the dot grew. This meant you could determine at a glance whose articles were shared most often.

Looking at the blue parts of the image, nothing was surprising: The largest circles were CNN and The New York Times, each shaded pale blue to indicate a center-left association. But the other side of the image showed just one big red circle: Breitbart. It was three times the size of Fox News and maybe a dozen times larger than any other news source on the right. If you wanted to know who was driving the Republican agenda in 2016, you didn't need to look much farther than the massive crimson orb parked on Benkler's screen.

Benkler sighed. If the Trayvon Martin study challenged his theory about the democratizing power of the internet, the Breitbart study offered an unsettling confirmation. ''An important part of what happened in this election is that a marginalized community, with views that were generally excluded, forced their way into the mainstream,'' he said. ''Now, whether that's 'democratizing' or not depends on how much emphasis you put on people being able to contest an election versus how much you put on civil rights, protection of minorities, rule of law. You could say that if it translates into denigration of minorities, it's antidemocratic. But as long as you are focused on the question of 'Do you have an intensely engaged minority able to clarify its message so clearly that it can contest politically in a way that it couldn't before?' then it's a democratizing effect.''

One of the interesting internet phenomena is that, whereas most of us can read the news generally without becoming disordered, the Left and Right can really only stand to read their own outlets, so every commenter from the bubbles sounds exactly the same and by glancing at a couple of their sites you can see what their daily talking points will be.

Posted by at August 16, 2017 7:42 AM