New Media Site Hopes to Provide Data-Driven "News about News"
BY Nick Judd | Tuesday, May 29 2012
It was April 2012 and although Newt Gingrich's presidential aspirations were fading away, his temper had not. He felt he had been treated unfairly and had no problems making that known.
"I think FOX has been for Romney all the way through," Gingrich was quoted as saying at the time. "In our experience, Callista and I both believe CNN is less biased than FOX this year. We are more likely to get neutral coverage out of CNN than we are of FOX, and we’re more likely to get distortion out of FOX. That’s just a fact."
Claims like these litter the campaign trail like candy wrappers, the discarded remnants of this or that attempt to control the message of the day. A new site, "The 4th Estate Project," hopes to join initiatives like the Pew Research Center's Project for Excellence in Journalism in quantifying the nebulous world of political journalism — where "narratives" drive the news and every poll has a margin of error — and in so doing provide more evidence to support or refute that type of claim. While tracking the tone and tenor of the news in aggregate isn't new, this project's business model is: The project's founders hope to start a media business driven by content a co-founder describes as "news about news."
"What I like about our focus on traditional media is you really are riding professional decisions," said co-founder Bryan Rich, a former Neiman fellow at Harvard. "You're basically quantifying highly professional, highly researched, highly codified behavior. And there's — in this country especially — a lot of epic pressure to be professional in terms of media. It may not always look that way, but that's why I think it's such a valid sample to analyze, because billions of billions of dollars are spent collecting, editing, improving, enriching the information and if we're able to quantify that then it's a high-value proposition in terms of the data."
Rich and developer Michael Howe originally put the underlying software to use at Rich's company, GlobalNI, which used it to provide intelligence for government organizations and big business. Rich and Howe say the original intent was to sell their data analysis to media companies, who might use it to track their own reporters' behavior; when that business didn't pan out, they found clients among Fortune 500 companies and in the military.
"We're capturing all the quotes in the media, so basically attributed statements and quotes, and we are then assigning metadata to those sources," Howe told me. "Anyone who is quoted is representative of a gender, a source type a political affiliation. And then we have these very interesting objects with these relationships between them. We've got the quotes, and the sources, and the metadata about the sources; we have the sentiment and topic information about the quotes; and then we've got the journalist who is using those sources, and we have the media connected with the journalist as well."
Claiming to be able to analyze the tone of a text is tricky, Howe admits. They use a human-assisted approach in an attempt to catch and fix errors.
They have released research breaking down news coverage of Gingrich by which corporate parent owned the outlet in question. (The Washington Post Company outstripped other comers when it came to apparently negative coverage of the former speaker, including Fox News, Howe found.)
Another post notes that only 18 percent of newsmakers quoted in print are female. A third says that a tally of all the attacks on Mitt Romney's personal life or character shows about 50 percent of those digs came from GOP contenders or their staff.
They're not the only people doing this kind of news analysis. In the wake of Gingrich's bias accusation, for example, Pew Research, using analysis tools from Crimson Hexagon, noted that coverage overall tended to note Mitt Romney more often, and more favorably, than Gingrich.
"It's hard to listen to Newt, right," Howe told me May 23, "so everyone just said, 'Yeah, right, whatever Newt,' and blew him off on that. But when we looked at the data it was very clear. It was clearer when we looked across News Corp and looked at that group as opposed to FOX," he said, but "there definitely was some pretty significant bias against Gingrich as opposed to Romney in the coverage by News Corp. To me that was interesting."
The entire company's origins, Rich told me, stem from a Nieman project he was involved with in 1997 to develop independent media in Burundi. To track bias and proper sourcing to a newsroom where reporters from the sharply divided Hutu and Tutsi ethnic groups were expected to work together, Rich said, he began tabulating the types of sources each journalist was using. Applying the same basic principle but using analysis software instead of hand tabulation and Excel spreadsheets, he says, he could now parse a reporter's clips file to reveal similar sourcing habits or editorial blind spots. More broadly, he sees a shot at a new business.
"What I see, the opportunity from a commercial point of view, is to create news about news that from my opinion is a direct public good because this should raise the game of the media, and it should make people aware of some of the gaps in stories as they're emerging," Rich said.