Listicles of “most outrageous 4chan comments,” presented so that readers will “understand what’s out there.” Nazi-next-door profiles like The New York Times’ infamous “A Voice of Hate in America’s Heartland.” Replications of offensive memes when a text description would suffice. Overly identifying information about harassment victims. Using the term “troll” when what you really mean is “violent misogynist or online stalker.”

These are just a few of the common practices that reporters and news outlets should avoid when they’re writing about extremists, antagonists, and manipulators online, writes Whitney Phillips in a Data & Society report released Tuesday. Phillips is an incoming assistant professor of communications at Syracuse University and author of the 2016 book This Is Why We Can’t Have Nice Things: Mapping the Relationship between Online Trolling and Mainstream Culture.

“The choices reporters and editors make about what to cover and how to cover it play a key part in regulating the amount of oxygen supplied to the falsehoods, antagonisms, and manipulations that threaten to overrun the contemporary media ecosystem — and, simultaneously, threaten to undermine democratic discourse more broadly,” Phillips writes. The “journalistic rule set” — traditional notions of “objectivity” and “both sides” reporting, for instance — must be reconsidered. For this paper, which was funded in part by the Craig Newmark Foundation and News Integrity Initiative, Phillips interviewed more than 50 people, including many reporters at online news outlets like BuzzFeed, Slate, and Motherboard; 56 percent of them were women, 30 percent were people of color, and 26 percent were natural-born citizens of countries outside the U.S.

The report is divided into three parts. The first offers a historical overview of the relationship between the news media and far-right manipulators during the 2016 election; here, Phillips focuses on young reporters who grew up online. The second part looks at “the intended and unintended consequences” of reporting on bad actors, as well as how journalism’s structural limitations can worsen outcomes. The third part offers some recommended practices and tips.

I found the first part of the report most fascinating, as it offers a theory of how a rise of young reporters hired to write about the Internet (often for Internet-only outlets) led to older reporters at legacy outlets reporting on the same issues, thus “resulting in a mushrooming of additional iterative coverage” and contributing to an increasingly emboldened group of white supremacists “who have shown themselves more than capable of taking their shitposting to the streets, with the Charlottesville white supremacist march being the most conspicuous, but hardly the only, example.”

Phillips distinguishes between young, “troll-trained” (or “on the internet”) journalists and (usually), “not troll-trained” journalists. Both of these groups have contributed to “catalyzing” the “unfolding alt-right narrative” in different ways. In the case of the younger respondents — 28- to 32-year-olds who work for Internet-focused publications or for the tech sections of long-established outlets, many of whom are quoted throughout this paper — Phillips found that “a reporter’s experience with trolling and/or 4chan strongly influenced how they initially approached stories about the alt-right and pro-Trump shitposting more broadly.”

This usually didn’t mean that the journalists themselves had spent a lot of time on 4chan or would have identified themselves as trolls, but “they verifiably aligned with 4chan’s symbolic demographics, particularly in terms of race,” Phillips notes, “all of the troll-adjacent reporters I spoke to, and all the reporters these reporters cited as further examples, are white.” “Forum kids,” one reporter called them. People who “thought the trolls on 4chan were nerds, and also worth getting into fights with.” Gawker reporters. While these young journalists’ relationships to 4chan varied, they were all “troll-trained” and “in a unique position to respond when pro-Trump rumblings first began emanating from sites like 4chan and Reddit, which many of these reporters had already been assigned as beats.” As Phillips tells it, based on her interviews, these reporters didn’t realize that something was different this time:

For those whose careers required them, daily, to plunge the internet depths, the memes, racist jokes, and general shitposting they were seeing at the outset of the election on Reddit and 4chan, as well as across their own Twitter and Facebook feeds, was entirely par for the internet course. These were the kinds of behaviors, and the kinds of people, they had been participating with, reporting on, and in many cases actively taunting, for years. They knew what to do. For the reporters “from the internet,” out came the listicles and other “weird internet” pieces that spotlighted the most outrageous and offensive memes circulating social media, which often affixed a shruggie-shaped question mark over whether the memes were “really” racist (as opposed to trollishly racist, which was treated as a different thing, per the presumably offset rules of the presumably offset weird internet)…For just about all of them, out came the Twitter snark about how “funny and bizarre” it was that “these people [were] using swastikas, using Nazi language to support Trump, as another former Gawker reporter explained…

Even as Trump inched toward the Republican nomination, many troll-trained journalists hadn’t yet realized that there were sincere neo-Nazis mixed in with the trolls…

No group was more remorseful than the reporters who applied weird internet framings or otherwise shined a half-righteous, half-ironic spotlight on early ‘alt-right’ antagonisms.

Meanwhile, Phillips writes, “other more traditional reporters” (as well as “reporters whose bodies numbered among those being targeted by far-right antagonists’ violent bigotries,” about whom Phillips writes more in the second section of the report) might have “[seen] the wolf” more clearly, but were also “particularly vulnerable to [the] subsequent onslaught of targeted manipulations.” In general, Phillips avoids calling out specific pieces as examples of egregiously bad reporting, but she does mention one in part because “Stein’s explicit framing of violent racism as being good for his story provides rare front-stage insight into an editorial calculus that typically only occurs backstage”: Joel Stein’s 2016 Time magazine cover story “How Trolls Are Ruining The Internet,” in which Stein got sucked into, and then printed in the piece, an email exchange with neo-Nazi and online abuser Andrew Auernheimer. “For a guy who doesn’t want to be interviewed for free, you’re giving me a lot of good quotes!” Stein wrote to Auernheimer — “good,” here, meaning “neo-Nazi hate mongering.”

“Without journalists reporting on them, there’s no way [far-right elements] would have gotten the attention they did,” Ashley Feinberg, now a senior reporter at HuffPost, told Phillips. “We’re setting the tone for them by covering them that way…at this point we have built the world they told us existed. We are the reason that these people are getting actual legitimate platforms now.” The Atlantic’s Emma Green, meanwhile, said, “It’s hard to delineate the bounds between what’s constructed and what actually would have existed without that kind of media attention.”

“Nothing,” writes Phillips, “has been better for alt-right trolling (whatever that word even means) than establishment journalism.”