Fake news: Duke researchers tally up ‘pink slime’ stories at Metric Media

By Ellen Clegg

You’ve no doubt heard of “pink slime.” To put it in polite terms, it’s the pastel-hued meat paste used in some processed hamburger meat. (I’m not linking to a photo: You can thank me later.)

Thanks to an episode of “This American Life,” it’s also a term that was used to describe Journatic, a company that outsourced hyperlocal “news” written by poorly paid offshore reporters — and marketed to mainstream newspapers. Ultimately, the fake bylines assigned to writers in the Philippines were a bridge too far for newspapers like the Chicago Tribune, so Journatic lost traction.

But the drive to try to automate local news — eliminating skilled reporters, assignment editors, copy editors, designers, photographers, product specialists, and engineers — hasn’t gone away, unfortunately. During the run-up to the 2020 presidential campaign, “pink slime” journalism came oozing back in the form of Metric Media, a vast network of “news” sites that aspire to a kind of quick-scan faux legitimacy. As Priyanjana Bengani reported in the Columbia Journalism Review, the number of Metric Media sites tripled over the course of 2020, constituting a shadowy network of supposedly hyperlocal outlets that were in fact funded by political operations and agenda-driven agencies and “designed to promote partisan talking points and collect user data.” Metric Media is run by Brian Timpone, and “is rooted in deception, eschewing hallmarks of news reporting like fairness and transparency,” according to an investigation by The New York Times. 

So how bad is Metric Media? Where’s the beef? Two professors at the DeWitt Wallace Center for Media & Democracy at Duke University have applied some actual metrics to Metric in a new study released this month, and the results are deeply troubling — especially against a backdrop of greed-fueled corporate media consolidation that has created ghost newspapers and news deserts across the country. Duke researchers Asa Royal and Philip M. Napoli scraped the front pages of 999 Metric Media outlets every day for more than two months to assess whether the content was serving communities or meeting an audience. The results of their study, “Local Journalism’s Possible Future: Metric Media and its Approach to Community Information Needs,” show glaring gaps in coverage and toxic levels of contempt for local readers. Some of their top-line findings:

• Most Metric Media sites lack any original content, and, in general, old content is not regularly updated. An overwhelming majority of stories appear to be automatically generated.
• In one 78-day observation period, nearly two-thirds of outlets did not publish a single article written by a human being.
• Stories about state and national politics are shown much more often than local news.
• Instead of democratizing local news for individual towns, Metric Media operates hubs in each state; on an average day, those hubs generate around 45% of the content shown across the network, despite only making up 5% of the network’s population.
• Metric Media dedicated an outsized amount of coverage to stories about electoral fraud in states where Donald Trump was contesting the vote during the 2020 general election.

Their conclusion: “Though the financial prospects for local newspapers are suffering, automated large-scale national operations appear, in this case, to be a poor substitute for the capital-intensive inputs of traditional local news.”

Royal and Napoli bring bracing gusts of data and fact to the ongoing effort to define what constitutes local news. And a reality check about the resource demands. When I was an editor at The Boston Globe, I was part of an effort to ramp up hyperlocal coverage in scores of cities and towns by creating twice-weekly print sections that were replated to zone in on targeted slices of the Globe’s coverage map: Globe West, Globe North, Globe NorthWest, Globe South. We hired reporters, editors, copy editors and advertising executives.

I spent a good part of a summer in the mezzanine-level production offices, the mailroom, and even on the loading dock on Morrissey Boulevard in Dorchester learning about the complex logistics of distribution and the timing of press runs. (Drivers knew the dimensions of their truck interior and calculated how many bundles of zoned sections would fit as they timed out their delivery route.) We experimented with different labels for the bundles and tested which were the most readable coming down the chute from the press. We leased office space in each region, bought scanners and copiers and computers.

Reporters attended town meetings and zoning board hearings and wrote everything from news stories to profiles to arts and entertainment features. Certainly, local government, schools, and police forces generated data that could be automated — data that readers wanted. And with a modern digital strategy, that data can be scraped and turned into news bites that are readily consumable, even welcome. But local doesn’t really scale — even if the cost structure of the rumbling, industrial print production cycle is stripped away for digital. Journalism that matters still means sustained investment in on-the-ground observation, in humans who can produce sustained beat reporting, painstaking investigations, pinpoint editing, and memorable visuals and design. Add in live events that allow for a robust exchange of ideas, even if only on Zoom, and advertising and marketing. “Pink slime” sites neglect a signal fact of human history, much in evidence during pandemic isolation: Connection and community are enduring. Storytelling is a timeless craft, which strengthens those bonds. In the end, that seems unlikely to be hijacked by an algorithm or a reporter-bot.