The U.S. media has come under intense scrutiny, with analysts, politicians, and even journalists themselves accusing it of bias and sensationalism — of having failed us — in its coverage of the presidential election. Critics across the political spectrum have said that fake news and cyberattacks played a big role in determining the course of events. The prevailing logic has an “if only” tenor: If only the media had been less swayed by shocking stories, if only bias in the media had been purged, and if only fake news had been eliminated and cyberattacks curtailed, the outcome would have been different. The presidential transition has been marked by the same attitude: if only the media were less distractible and headlines more accurate.
Thinking that way is tempting, but it misses the mark. The media did exactly what it was designed to do, given the incentives that govern it. It’s not that the media sets out to be sensationalist; its business model leads it in that direction. Charges of bias don’t make the bias real; it often lies in the eye of the beholder. Fake news and cyberattacks are triggers, not causes. The issues that confront us are structural.
To the question, If the media were to cover the election again, with the benefit of hindsight, could we expect anything different? my answer is a sobering no. This is for two reasons: the way news is produced and amplified (the supply side) and the way consumers process news (the demand side).
A caveat is in order. The analysis here is not concerned with which candidate deserved to win or whose message was “better.” It is concerned with examining the media and its coverage, identifying its root causes, and understanding what we should expect going forward.
The Supply Side I: Connectedness Matters More than Content or Money
Political campaigns are marketing campaigns, messages aimed at selling a product. Like marketers, politicians obsess over messaging (what journalists would call “content”) and a few key metrics that historically have determined success: amount of television advertising, number of “foot soldiers,” intensity of get-out-the-vote operations, and voter demographics. But in the last two contests in which Hillary Clinton has participated, the 2008 primary and the 2016 election, she won on most of these metrics — and lost the elections.
Two developments bear noting. First, and most obvious, traditional media is no longer the only way to spread the word. Any candidate can communicate directly and instantly with millions of people. Media companies are experiencing an extreme form of competition that comes with digital technologies: Everyone is a media company today.
Second, and even more significant, social media is distinct from traditional media in that it connects users to each other. This means that messages can spread far more easily and quickly (compare how often you share a TV ad and a tweet).
Essential Background
The implications are threefold:
The best product doesn’t always win. Even if you have the best product or candidate, if you run a hub-and-spokes campaign, you’ll attract followers one by one. Create a product or candidate that connects users, and your message — and advantage — will spread rapidly. Apple learned this the hard way. For 20 years, starting in 1984, the Macintosh was superior to any PC. Yet by 2004 its market share was down to 3%. Apple had a great product, but Microsoft had a network of connected users. Because more people used PCs, and wrote software for them, they became the default choice for nearly everyone.
Many organizations and entrepreneurs miss this lesson. Focus only on creating the best content or product, and you can lose because of untapped user connections — a phenomenon I call the “content trap.” It explains why firms that have anchored their strategies to content have ceded digital leadership to those that have focused on connections.
Consider the Scandinavian media firm Schibsted, which engineered an impressive digital transformation through a philosophy of connectedness. It focused its efforts on earning a majority share of Europe’s digital classified advertising market (a product that connects buyers and sellers). It then shifted its news focus from great content to content rooted in the question “Can we help readers help each other?” During the volcanic ash crisis of 2010, what it offered wasn’t prize-winning stories about the roots of the eruption or its health implications, but an app (Hitchhiker’s Central) that allowed readers to share travel plans and offer rides to each other. Similarly, during the 2016 election, many American voters found journalistic content less relevant than what they were experiencing in their own lives.
Bigger marketing budgets may not pay off. In a digital world full of product clutter, the best marketing campaigns spend nearly nothing. JC Penney spent no money on television advertising during the 2015 Super Bowl, yet its “mittens” campaign was one of the most watched. The campaign relied solely on Twitter and went viral by virtue of intentional spelling mistakes. Once a “connected” product draws in users, those users effectively become the sales force. Facebook, Uber, and Airbnb are all examples of this. Donald Trump spent only half of what Clinton did during the campaign.
Expectations matter. In connected worlds, expectations about future growth affect what current users choose; people want to be on a winning platform. This has led to a strategy known as vaporware, a term for when firms announce strengths they may not possess or supposedly imminent product launches to draw users. Consider Trump’s first words in the June 2015 announcement of his candidacy: “Wow. Whoa. That is some group of people. Thousands.…This is beyond anybody’s expectations. There’s been no crowd like this.” This wasn’t just a campaign message; it was an effort to shape expectations and trigger connectedness.
The Supply Side II: Ratings Determine Which Messages Get Amplified
The first phase of a marketing campaign is deciding how and where to spend your marketing dollars. The second is influencing how your message gets amplified. One of the most important mechanisms for this is traditional media — so-called “earned media coverage.” You can spend a lot in the first phase and get little amplification in the second, or vice versa.
Recycling the same message won’t earn amplification. And in today’s media environment, even “normal” news doesn’t break through information clutter; big, surprising events do. The media’s bias toward big events stems from three features of its economics:
Fixed costs. The cost of covering a golf tournament doesn’t depend on whether Tiger Woods plays. But if he does, ratings — and revenue — double. The same phenomenon affects decisions about covering news stories or political rallies.
An advertising-based model. Advertising (and other indirect charges like cable operator fees) are central to the economics of most news media, and this creates a bias whereby the number of viewers is more important than whether viewers like the coverage. (What matters is that you watch news coverage, not whether you are ready to throw a chair at it out of disgust.) Fixed costs have always been central to the economics of media. Advertising came later — and when it did, in the early 20th century, news became more sensational. That’s hardly surprising: The main metric by which news outlets are judged is the ratings they command, the page views they get, or the copies they sell.
Spillovers. A big event in media and entertainment doesn’t just draw viewers to the event itself; it also entices viewers to consume follow-on or related products (and a company’s previous products, too). People who watch a television program are far more likely to watch the next program on that channel, for example.
Each of these factors, individually, means that ratings or page views — the size of the audience — matter a lot for media firms. Together, they lead to a fixation on ratings to the exclusion of almost anything else. Competition further reinforces this dynamic, making audience size the metric by which media firms are measured. The outcome is a “ratings bubble” within which companies operate.
Big-event bias is even more pronounced in entertainment worlds, where getting noticed has gotten increasingly hard over time. This explains the trend toward spinoffs, sequels, and franchises in broadcast television and movies (viewers are already familiar with the basic story) and big-name authors in books (they generate publicity) and why successful sports franchises tend to get even more successful over time (they draw lots of viewers, which allows them to spend more on star players, who draw even more viewers). Success might have more to do with awareness than with quality. When the pseudonymous Robert Galbraith published A Cuckoo’s Calling in 2013, the novel sold about 1,500 copies in the first month. After the author was revealed to be Harry Potter creator J.K. Rowling, sales rose to over one million.
Piggybacking on big events has allowed certain media companies to grow over time. Fox News, for instance, entered the seemingly mature cable market in 1996 and experienced notable upticks in viewers after “big news” events — the 2000 election, the 9/11 terrorist attacks, and the start of the war in Iraq. When an event drew viewers to cable news in general, Fox’s ratings grew along with the other networks’. But more of the viewers who tuned into Fox stayed with it after the event had passed when they realized the network’s coverage was different.
In political campaigns, big events arise in one of three ways. The first is sporadically and unpredictably, as with the San Bernardino shooting or the Access Hollywood tape. The timing of such surprises can be particularly fortuitous or damaging (see: James Comey). The second is through name recognition. Events become more newsworthy if they’re accompanied by a big name. The third is by being created. Steve Jobs understood this more than most technology executives, which is why he elevated product launches to an art form: Every media firm had to cover a new Apple release. And Trump understood this more than any other candidate: Every time he made a provocative comment on a new subject, the news outlets covered it.
These forces help explain why Trump got so much more media coverage than, say, Bernie Sanders, who touted a similarly antiestablishment, populist message. Populism and inequality aren’t news; calling Mexican immigrants rapists and vowing to build a wall are. So Sanders’s brand of populism wasn’t news; Trump’s was. The reason was rooted in media economics, not in the effort or preferences of journalists and programming executives. A combination of fixed costs, an advertising-reliant model, and spillovers produced a staggering difference in earned media coverage during the primaries: $2 billion for Trump and $300 million for Sanders. Television advertising, where Clinton had a huge leg up on both, hardly seemed to matter at all.
Competition Can Backfire
Competition and private firms operating in their self-interest typically lead to well-functioning markets. But that’s not always what happens. A well-known exception occurs when externalities exist — side effects on other people or firms that aren’t usually accounted for by private actors. (Canonical examples are cigarette smoking or pollution, or a store manager in a large retail chain pursuing actions that benefit his individual store but damage the parent company’s brand.) In situations like these, following your self-interest (in this case, as a media firm) doesn’t necessarily further the collective good, or even your own.
In 2009 Netflix needed high-quality content to grow its streaming business. It could get that content only from Hollywood studios. The studios had seen Netflix grow its DVD business for a decade, and now, with a stronger bargaining position in the streaming market — the first-sale doctrine that allowed any DVD owner to resell did not apply to streaming — they could have chosen not to license to Netflix and nipped it in the bud. But they granted licenses, and Netflix soon became the giant they hadn’t wanted to see arise. Why did the studios act against their own interests?
If they could have collectively agreed not to license to Netflix, the result would have been different. But they couldn’t. At first only Viacom relented, licensing archived Beavis and Butt-head episodes. One show, it reasoned, could not a streaming giant make. But then everyone followed that logic.
It wasn’t that the content providers didn’t see what was happening; it was that they couldn’t coordinate. It’s why newspapers let Google crawl their content for Google News. It’s why they handed content to Facebook for its Instant Articles format last year.
So, too, with the recent political campaign. If every media outlet had ignored Trump’s rallies and rhetoric, it would have paid handsomely for one outlet to cover them. But once one did cover them, no others could afford not to.
These events coalesced dramatically toward the end of the campaign, when Trump announced a press conference in which he would ostensibly make a major announcement about President Obama’s birth certificate (a lie that he had prolonged that had found traction in media coverage several years back). Nearly every media outlet showed up. How could they not cover a major announcement by a presidential candidate? But it was a sham — there was no real announcement, other than that there would be no more announcements on the subject.
This is the prisoner’s dilemma of reporting amid competition: Following your self-interest does not always further the collective good. The situation generated one of the most dispiritingly candid statements ever from a media executive: Early in 2016, when the head of CBS was asked about the disproportionate attention given to Trump, he quipped, “It may not be good for America, but it’s damn good for CBS.”
The network wasn’t alone. Cable news outlets enjoyed similar gains in 2016, marking it as their best year ever. Meanwhile, public trust in the press reached its lowest level in history.
The Demand Side: Consumers Consume What They Want To
One of the longest-standing debates in marketing is not whether advertising works, but how it does. One view is that marketing persuades consumers to purchase. Hear a song once, and you may not like it; hear it repeatedly, and you’ll start to, regardless of how good or bad it is (hence the phrase “all publicity is good publicity”). Others argue that marketing merely increases awareness without altering beliefs. By this reasoning, repeated exposure to a song that doesn’t match your taste might make you less likely to buy it.
Does media reporting change what we believe, or do our preferences shape what media we choose to watch in the first place?
Most research indicates that the latter is central: Our preexisting preferences largely determine what media we watch. One of the most reliable findings in the study of television entertainment is that viewers watch programs whose characters are like themselves. Older people watch shows featuring older characters, younger viewers watch shows featuring younger ones; the same goes for gender, ethnicity, and income. A similar effect is seen in news: We watch outlets whose reporting is consistent with our beliefs. Viewers who identify with the right are more likely to watch Fox, while left-leaning people are more likely to watch MSNBC. Similar differences apply to intra-network program choices, since programs on the same network can differ in their positioning.
These patterns in news-watching would be puzzling if all that news providers did was provide verifiably objective information. But like entertainment programs, news programs and channels differ in their positioning, in the way they report information (often referred to as slant), and in what information they report (agenda setting). News positioning matters — viewers watch news programs and channels whose positions match their tastes and beliefs.
This pattern of sorting on beliefs is amplified over time by various additional factors. The first is competition among media, which has increased as digital technologies have led to a vast number of new media outlets, each catering to more-niche tastes. The second is viewers’ confirmation bias, which leads us to reject valid information that is not consistent with our beliefs. Confirmation bias is deeply rooted in human behavior. It affects not just how we process information but who we associate with, creating “filter bubbles.” These bubbles are further reinforced by website algorithms designed to personalize the information we receive based on our past behaviors. Persuasive effects of the media also serve to solidify these bubbles. (And even small persuasive effects can have large effects in close elections.)
Each factor increases viewer polarization, which on certain measures has reached unprecedented levels. Together, they shape how we respond to bias in the media. Consider the debate over left and right media bias, which goes back several decades and has grown in intensity over time. Part of what makes discussions of bias so thorny is that we almost never agree on what bias is. Both the debate and studies tend to focus on what the media reports — on content. But studies show that content is not the only place where bias lives. In experiments, when two people with different beliefs view exactly the same content, their perceptions of bias differ.
Add it all up, and the implications are profound.
First, we watch what we believe, but what we don’t watch, we don’t believe. This is the effect of sorting based on beliefs.
Second, negative coverage can have unintended consequences. Hear a source you don’t trust, and when it reports something inconsistent with your beliefs, you’ll discount that thing even more. (The rare exception is when events are incontrovertibly verifiable — for example, the question of who said what on the Access Hollywood tape.) During the election season, more newspapers endorsed Clinton than any presidential candidate in U.S history. Papers with a tradition of endorsing Republicans endorsed her; papers with a tradition of not endorsing a candidate did, too. But none of it mattered; editorial content was essentially irrelevant.
Third, and for the same reason, charges of media bias can actually help an outlet. The more your favorite channel is alleged to be biased by people you disagree with, the more you’ll watch it. Trump wasn’t the first to see this phenomenon: In Fox News’s early days, senior executives often acknowledged that charges of bias appeared to help them. And it isn’t specific to right-leaning voters. After the election, when Trump tweeted complaints about the New York Times and Vanity Fair, both outlets saw a rise in subscriptions. Charges of bias harden beliefs and reinforce polarization.
Particularly sobering is that all this has nothing do with the much-lamented problem of fake news. Get rid of all verifiably fake news, as Facebook and others certainly should, and filter bubbles, polarization, and charges of media bias will remain.
Where Does This Leave Us?
Three forces combine to create the media coverage of political campaigns we observe today: connected media, which spreads messages faster than traditional media; fixed costs and advertising-reliant business models in traditional media, which amplify sensational messages; and viewers’ news consumption patterns, which leads to people sorting across media outlets based on their beliefs and makes messages they already agree with far more effective. Each reinforces the others. Without these enabling factors, even the best marketing campaign would go nowhere, and fake news or leaked information from cyberattacks would have little effect.
Fair questions have been raised about the lack of investigative journalism early in the campaign, false equivalencies in reporting, and the use of paid campaign operatives as experts on television news. But digital technology and business incentives exerted more influence over the media coverage than editorial decisions and missing voices did. The ratings bubble had as much impact as filter bubbles did. The forces at work here — the search for profitability, competition, and self-interest — are things we embrace as profoundly American.
Competition in the media leads to efficiency as well as to checks and balances — all good things. But it fails to internalize the externalities from profitable but sensational coverage. It leads to differentiation and more voices (also good, and what’s been the focus of regulatory efforts) but also to fragmentation, polarization, and less-penetrable filter bubbles (dangerous).
It’s tempting to stretch the analysis between marketing and politics too far. They are different in important respects. Most notable, in marketing you can win through strategies that exploit the big-event bias of media (through attention-grabbing rhetoric) and the beliefs of consumers (through allegations that discredit your competitors). These strategies draw in consumers who are right for your brand. But in presidential politics, the same approach is incredibly risky because when you win, you serve everyone, not just those who “purchased your product.” Despite these differences, the same economics of information supply-and-demand that shape digital strategies in business are doing so in politics.
Which leads to my conclusion: Even if we could somehow push “reset,” we would have to expect the same sort of coverage that we got. The problems are too deep and structural for anything else.
What’s the way forward? There are no easy answers to the question. This analysis mainly points to solutions that won’t work. Voluntary efforts at restraint by well-meaning journalists won’t work, because of advertising-based business models and competition. Eliminating fake news won’t change the fact that voters ignore ideas contrary to their beliefs. And it won’t solve the media’s structural challenges or change its incentives. Media companies, their regulators, and their customers — all of us — have to look for ways to confront these challenges. The stakes could not be higher.
