*(I originally wrote this before the COVID-19 pandemic hit the United States, and since then, the pandemic has further exposed the depth of this problem. As a result of the problems described herein — most notably, advertiser keyword blocking of “coronavirus” and “COVID-19” — reputable news publishers are suffering terribly from loss of advertising revenue).
With today’s enormous number of news (and news-like) sources, which vary greatly in degrees of reliability and bias, one can’t help but be interested in two questions: 1) how much have news sources increased in number over time, and 2) what caused that increase?
The answer to the first question is difficult to pin down exactly and numerically, but easy to discern in more general terms just by having observed the world over the last 20 to 70 years. News sources have grown in number a lot — at first incrementally, and then exponentially and overwhelmingly.
What’s immediately knowable is that in the 1960s and ’70s, Americans got their news from local and national newspapers and just three major broadcast networks: ABC, NBC, CBS; in the 1980s and 90s, 24-hour cable news — including the partisan kind — became available; and in the 2000s, news moved online. Then in the 2010s, with the growth of Facebook, Twitter, and YouTube, the number of news sources exploded.
The answer to the second question — what caused that increase — is really several answers, many of which are also immediately knowable and seemingly obvious. Technology advances, especially those related to the growth of the internet and the rise of social media platforms, dramatically lowered the ease with which new news sources could emerge, and the ease with which they could widely distribute their content. What used to cost enormous investments in financial and human capital for the generation and distribution of news-like content can now be done at a fraction of the time and cost; nearly anyone can set up a news website, create content, and get distribution for less than a few hundred dollars, in as little as a few days.
This lowering of the barrier to entry enables growth, but an important incentive is also at play: the ability to generate revenue from creating news content. Technology has fueled this ability for nearly anyone to monetize content, but it wouldn’t be possible without the providers of the revenue themselves: the advertisers.
Unfortunately, digital advertising models have created perverse incentives for certain creators of news-content — namely, junk news providers. All news content creators that are competing for attention in a brutally competitive market need to drive traffic and engagement to maximize their revenue potential.
The challenge is different for news sources at the top of the Media Bias Chart® than for those at the bottom. For long-existing, reputable news organizations at the top, the competition has required reworking of business models that include advertising, but they are somewhat constrained (thankfully) in the changes to their content by the need to uphold journalistic standards for reliability to maintain their reputations and keep large readerships.
Newer, digital-only entrants to the news content space are not constrained in the way established publishers are. A small, online operation run by one person or a handful of people can carve out a space for itself that generates plenty of money to sustain and grow it by generating web traffic and social media engagement. Unfortunately, some of the easiest ways to generate those are by publishing news content that is hyper-partisan, extremist, misleading, sensationalistic, propagandistic, or downright false. News content that drives fear, anger, disgust, and righteousness about one’s own side is highly attracting and engaging. It can generate millions of page views on an individual site, and millions of page views can translate into tens of thousands of dollars per month or more in digital advertising revenue. In other words, polarizing junk news pays.
Since the dawn of digital advertising, the focus of advertising technology has been on targeting the brand’s most likely demographics, and as it improved, its most likely consumers, such that now, ad targeting utilizing the personal information of the individual user is the norm. Digital ads target the user above all else, using cookies, location information, and other data pieces about the individual. Usually this targeting is done with little or no knowledge (or concern) by the brand about the content on which the advertisement is placed. Whether the ad appears online in USA Today (a good journalism source) or The Gateway Pundit (a junk news source) has almost everything to do with targeting the user and almost nothing to do with the content on those sites. Most of this ad purchasing is done automatically (i.e., “programmatically”).
As a result, brands are funneling money to both reputable journalism outlets and polarizing junk news outlets in pursuit of their target audience; both good and bad publishers therefore get rewarded for attracting that target audience, no matter how they do it. Current tools and technology make it hard for brands to distinguish whether a site is attracting an audience by performing dogged investigative reporting of an important issue to the world, or by whipping gullible people into a frenzy of anger and division with falsehoods. Both kinds of sites are still attractive to advertisers because of the reach into their potential audience.
Only recently have advertisers started to realize that the news content by which they advertise can be damaging (to themselves). At first it was a realization due to brand-specific embarrassing placements; a burger ad next to an E.coli outbreak story; an airline ad next to a plane crash story. Then, a broader realization that no brand should advertise, and thereby support, content that comprises pornography, violence, or hate speech led to the need for “brand safety” precautions. Some research has shown that consumers often believe brands intentionally choose where they place their ads (even though that is usually not the case). Even when consumers know brands don’t control ad placement, most still feel that brands should be responsible for where they place them. The term “brand safety” itself implies that this is more for the good and reputation of the brand, rather than the demonetization of the content out of concern for the consumer. Of course, this kind of content (especially porn) can monetize itself without ads, so it’s not as if advertisers are responsible for the very existence of this content.
News outlets (both good and junk ones), however, have a very difficult time monetizing without ads. The possible revenue models for news outlets are 1) subscriptions, 2) grants/donations, or 3) advertising, or a combination of the three. At the top of the Media Bias Chart, good journalism outlets have suffered because of reductions in advertising revenue due to multiple factors, including competition, leading to the cutting of newsroom staff. This suffering has been felt most acutely by local news outlets, to the well-documented detriment of many communities
The sources at the bottom of the Media Bias Chart have benefited greatly from the lack of care by brands that has resulted in revenue flowing to their pages — the pages that rip at people’s basest instincts for fear and hatred of others and exacerbate the partisan divisions that plague our country. Junk news plays a significant role in our national polarization (see, e.g., Ezra Klein’s recent book, Why We’re Polarized). The danger of polarization of news media is unfolding tragically into a partisan divide over how to respond to the coronavirus outbreak, with many right-wing news sources purposefully downplaying it for perceived political benefit. Center-left and left-wing sources that are always quick to emphasize or exaggerate Trump’s shortcomings had long since lost credibility with right-leaning audiences such that their factual reporting about the virus was distrusted by such audiences from the outset. As a result, Republicans and Democrats believe dangerously different things about the risks of the outbreak, resulting in many Republicans generally taking the outbreak less seriously. It is becoming apparent that polarizing junk news is costing people’s lives.
Consequently, it is more important than ever to address why polarizing junk news sources exist in the first place. The overwhelming majority of these side- and bottom-dwellers of the news landscape do not rely on subscriptions or donations. They are almost all driven by advertising revenue from brands big and small. It is safe to say that without brands’ advertising dollars, many of these sites wouldn’t exist. Amazon wouldn’t pay a dollar to a man on a street corner shouting conspiracy theories into a megaphone to hold an Amazon sign while he does it; but put this same man and his conspiracy theories on a website that gets a few hundred thousand clicks per month, and Amazon will pay him many dollars. So will Century Link, Nordstrom, HBO, and many more.
Some brands do care that their ads might appear next to news content they deem not to be “brand suitable.” However, the term “brand suitable” is often poorly defined by brands themselves when it comes to determining what news content is “suitable.” For many brands, the term broadly means that they want to avoid advertising next to content that is “negative” or “political.” What brands should care about is that they not be advertising next to news content that is too unreliable or too biased — content we refer to as “polarizing junk news.”
To the extent that some brands are starting to care about the fact that their ads appear next to polarizing junk news, they don’t have the tools to avoid it without avoiding good journalism at the same time. The one way for a brand to ensure they buy ads on good journalism sources is to directly buy them from the publisher, (e.g., Toyota’s agency would go and buys ads directly from the New York Times), but those types of transactions comprise just a small percentage of digital advertising on news sources. Most of this advertising is done programmatically (through a maze of digital advertising tools which buy and sell ads on sites instantaneously).
These ad tech tools provide an array of filters and criteria by which to target individual consumers; there is abundant technology for brands to make choices on this aspect of their ad buys. There are also some brand safety filters that help advertisers avoid pornography, violence, and hate speech content. These typically comprise keyword filters that can scan URLs and web pages for explicit language, racial slurs, and other generally offensive terms, and blacklists of previously known unacceptable websites.
No useful tools exist for advertisers to distinguish and choose between ad buys on good journalism and junk news. The tools that exist are limited to these same keyword filters and blacklists. But these tools are far too blunt to be useful. If an advertiser wants to avoid polarizing junk news, the main options are to use keywords commonly used in polarizing news stories and block them — but what are those words, exactly? The ones tried on some blacklists include “LGBT,” “shooting,” “Trump,” and most recently “coronavirus” and “COVID-19.” One can easily see how that would result in the exclusion of important stories on reputable sites. Some advertisers choose an even more blunt approach: to avoid any news content altogether.
Good journalism outlets are acutely aware of how they have been hurt by both keyword blocking and news content avoidance. An article in the Wall Street Journal in August of 2019 stated, “Some companies are creating keyword blacklists so detailed as to make almost all political or hard-news stories off-limits for their ads.” The recent pandemic has turned this existing problem into a crisis, with keyword blocking of coronavirus stories starving good publishers of vital revenue at the very moment they are doing the most crucial work.
The Drum, an ad industry publication, stated that advertisers are in a “race to brand safety” that “the current reliance upon keywords is disproportionately (and ironically) penalizing, sites with a greater journalistic impact who run high quality, editorially-governed environments.” The article highlighted a recent study commissioned by cybersecurity firm CHEQ that estimates “top media owners are being penalized with blunt brand-safety tools to the tune of $3.2bn a year across the US, UK, Japan and Australia.”
The inability for advertisers to distinguish between good journalism and junk news, therefore, causes a vicious cycle that worsens our entire news ecosystem. Brands that care penalize good journalism; brands that don’t care reward junk news creators.
This problem is solvable, though. Brands need to go beyond the brand safety and brand suitability paradigms and consider their “brand conscience” regarding their role in the news. To act according to their brand consciences, they need a new screening tool. A comprehensive solution requires news content ratings for reliability and bias, like the news content ratings created by Ad Fontes Media. Such ratings need to be used in place of keyword blacklists in the ad tech system. Ratings need to account for the entire range of reliability — from the best sources down to the worst sources, and the range of bias, from those with minimal bias to those with the most extreme types of bias.
Such ratings can identify junk news. These ratings need to be done at the source level and at the article level. They need to be numeric and have thresholds. They need to be non-partisan and they need to be accurate. In order to be accurate, news content ratings need to be done with a combination of human review and automated tools, not inaccurate technological shortcuts.
News content ratings make responsible news advertising possible. But it will first require news consumers, publishers, and brands to recognize the role ads have played in making our news ecosystem worse and demanding that brands work to make it better.
 The term “news-like” sources refers to any kind content from which people derive some kind of information that could broadly be considered to be news, with the implication that many of these sources are undesirable in terms of reliability and/or bias. Whenever the article refers to “news sources” or “news content,” it should be understood that this includes “news-like” sources or content.