Trust Me, I'm Lying: Confessions of a Media Manipulator

by Ryan Holiday

  ONE SPREADS, THE OTHER DOESN’T

  In an article in the New Republic called “The Case Against Economic Disaster Porn,” Noreen Malone points out that one thing stands out about the incredibly viral photographs of Detroit: Not a single one of the popular photos of the ruins of Detroit has a person in it. That was the difference between the Huffington Post slideshows and the Magnum photos—Magnum dared to include human beings in their photos of Detroit. The photos that spread, on the other hand, are deliberately devoid of any sign of life.2

  Detroit has a homeless population of nearly twenty thousand, and in 2011, city funding for homeless shelters was cut in half. Thousands more live in foreclosed houses and buildings without electricity or heat, the very same structures in the pictures. These photos don’t just omit people. Detroit is a city overrun by stray dogs, which roam the city in packs hunting and scavenging for food. Conservative figures estimate that there are as many as 50,000 wild dogs living in Detroit and something like 650,000 feral cats. In other words, you can’t walk a block in Detroit without seeing heartbreaking and deeply wounded signs of life.

  You’d have to try not to. And that’s exactly what these slideshow photographers do. Why? Because all that is depressing. As Jonah Peretti, the virality expert behind both the Huffington Post and BuzzFeed, believes, “if something is a total bummer, people don’t share it.” And since people wouldn’t share it, blogs won’t publish it. Seeing the homeless and drug addicts and starving, dying animals would take away all the fun.* It’d make the viewers feel uncomfortable, and unsettling images are not conducive to sharing. Why, Peretti asks, would anyone—bloggers or readers—want to pass along bad feelings?3

  The economics of the web make it impossible to portray the complex situation in Detroit accurately. It turns out that photos of Detroit that spread do so precisely because they are dead. Simple narratives like the haunting ruins of a city spread and live, while complicated ones like a city filled with real people who desperately need help don’t.

  One city. Two possible portrayals. One is a bummer, one looks cool. Only one makes it into the Huffington Post slideshow. Only one is worth trying to sell the bloggers.

  THE DNA OF THE VIRUS

  Only a certain style of video, article, or tweet has the ability to rise above the overwhelming noise and make an impression. But the web is not some fair or positive meritocracy, and the first comprehensive study on why this is bears this out. In 2010, two researchers at the Wharton School looked at seven thousand articles that made it onto the New York Times Most E-mailed List. (A story from the Times is shared on Twitter once every four seconds, making the list one of the biggest media platforms on the web.) The researchers’ results confirm almost everything we see when content like the sensational ruin porn of Detroit goes viral. For me it confirmed every intuition behind my manipulations.4

  According to the story, “the most powerful predictor of virality is how much anger an article evokes” [emphasis mine]. I will say it again: The most powerful predictor of what spreads online is anger. No wonder the outrage I created for Tucker’s movie worked so well. Anger has such a profound effect that one standard deviation increase in the anger rating of an article is the equivalent of spending an additional three hours as the lead story on the front page of NYTimes.com.

  Again, extremes in any direction have a large impact on how something will spread, but certain emotions do better than others. For instance, an equal shift in the positivity of an article is the equivalent of spending about 1.2 hours as the lead story. It’s a significant but clear difference. The angrier an article makes the reader, the better.

  The researchers found that while sadness is an extreme emotion, it is a wholly unviral one. Sadness, like what one might feel to see a stray dog shivering for warmth or a homeless man begging for money, is typically a low-arousal emotion. Sadness depresses our impulse for social sharing. It’s why nobody wanted to share the Magnum photos but gladly shared the ones on the Huffington Post. The HuffPo photos were awe-some; they made us angry, or they surprised us. Such emotions trigger a desire to act—they are arousing—and that is exactly the reaction a publisher hopes to exploit.

  In turn, it’s what marketers exploit as well. A powerful predictor of whether content will spread online is valence, or the degree of positive or negative emotion a person is made to feel. Both extremes are more desirable than anything in the middle. Regardless of the topic, the more an article makes someone feel good or bad, the more likely it is to make the Most E-mailed list. No marketer is ever going to push something with the stink of reasonableness, complexity, or mixed emotions.

  Yet information is rarely clearly good or bad. It tends to have elements of both, or none of either. It just is. Navigating this quandary forces marketers and publishers to conspire to distort this information into something that will register on the emotional spectrum of the audience. To turn it into something that spreads and to drive clicks. Behind the scenes I work to crank up the valence of articles, relying on scandal, conflict, triviality, titillation, and dogmatism. Whatever will ensure transmission.

  The media is in the evil position of needing to go negative and play tricks with your psyche in order to drive you to share their material online. For instance, in studies where subjects are shown negative video footage (war, an airplane crash, an execution, a natural disaster), they become more aroused, can better recall what happened, pay more attention, and engage more cognitive resources to consume the media than nonnegative footage.5 That’s the kind of stuff that will make you hit “share this.” They push your buttons so you’ll press theirs.

  Things must be negative but not too negative. Hopelessness, despair—these drive us to do nothing. Pity, empathy—those drive us to do something, like get up from our computers to act. But anger, fear, excitement, or laughter—these drive us to spread. They drive us to do something that makes us feel as if we are doing something, when in reality we are only contributing to what is probably a superficial and utterly meaningless conversation. Online games and apps operate on the same principles and exploit the same impulses: be consuming without frustrating, manipulative without revealing the strings.

  For those who know what levers provoke people to share, media manipulation becomes simply a matter of packaging and presentation. All it takes is the right frame, the right angle, and millions of readers will willingly send your idea or image or ad to their friends, family, and coworkers on your behalf. Bloggers know this, and want it badly. If I can hand them a story that may be able to deliver, who are they to refuse?

  GIVING THE BASTARDS WHAT THEY WANT

  When I design online ads for American Apparel, I almost always look for an angle that will provoke. Outrage, self-righteousness, and titillation all work equally well. Naturally, the sexy ones are probably those you remember most, but the formula worked for all types of images. Photos of kids dressed up like adults, dogs wearing clothes, ad copy that didn’t make any sense—all high-valence, viral images. If I could generate a reaction, I could propel the ad from being something I had to pay for people to see (by buying ad inventory) to something people would gladly post on the front page of their highly trafficked websites.

  I once ran a series of completely nude (not safe for work, or NSFW) advertisements featuring the porn star Sasha Grey on two blogs. They were very small websites, and the total cost of the ads was only twelve hundred dollars. A naked woman with visible pubic hair + a major U.S. retailer + blogs = a massive online story.

  The ads were picked up online by Nerve, BuzzFeed, Fast Company, Jezebel, Refinery29, NBC New York, Fleshbot, the Portland Mercury, and many others. They eventually made it into print as far away as Rolling Stone Brazil, and they’re still being passed around online. The idea wasn’t ever to sell products directly through the ads themselves, since the model wasn’t really wearing any of it—and the sites were too small, anyway. I knew that just the notion of a company running pornographic advertisements on legitimate blogs would be t
oo arousing (no pun intended) for share-hungry sites and readers to resist. I’m not sure if I was the first person to ever do this, but I certainly told reporters I was. Some blogs wrote about it in anger, some wrote about it in disgust, and others loved it and wanted more. The important part was that they wrote about it at all. It ended up being seen millions of times, and almost none of those views was on the original sites where we paid for the ads to run.

  I wasn’t trying to create controversy for the sake of controversy. The publicity from the spectacle generated tens of thousands of dollars in sales, and that was my intention all along. I had substantial data to back up the fact that chatter correlated with a spike in purchases of whatever product was the subject of the conversation. Armed with this information, I made it my strategy to manufacture chatter by exploiting emotions of high valence: arousal and indignation. I’d serve ads in direct violation of the standards of publishers and ad networks, knowing that while they’d inevitably be pulled, the ads would generate all sorts of brand awareness in the few minutes users saw them. A slight slap on the wrist or pissing off some prudes was a penalty well worth paying for, for all the attention and money we got.

  In the case of American Apparel, this leveraged advertising strategy I developed was responsible for taking online sales from forty million dollars to nearly sixty million dollars in three years—with a minuscule ad budget.

  HIDDEN CONSEQUENCES

  I use these tactics to sell products, and they work—lots of product gets sold. But I have come to know that the act of constantly provoking and fooling people has a larger cost. Nor am I the only one doing it.

  You probably don’t remember what happened on February 19, 2009, and that’s because nothing notable happened—at least by any normal standard. But to those who make their living by “what spreads,” it was an incredibly lucrative day, and for our country, it was a costly one.

  During what was supposed to be a standard on-camera segment, CNBC correspondent Rick Santelli had a somewhat awkward meltdown on the floor of the Chicago Mercantile Exchange. He went off script and started ranting about the Obama administration and the then recently passed stimulus bill. Then he started yelling about homeowners who bit off bigger mortgages than they could chew, and Cuba, and a bunch of other ridiculous stuff. Traders on the floor began to cheer (and jeer), and he ended by declaring that he was thinking about having a “Chicago Tea Party” to dump derivatives into Lake Michigan. The whole thing looked like a shit show.

  CNBC was smart. They recognized from the reaction of their anchors—which ranged from horror to mild bemusement—that they had something valuable on their hands. Instead of waiting for the video to be discovered by bloggers, news junkies, message boards, and mash-up artists, CNBC posted it on their own website immediately. While this might seem like a strange move for a serious media outlet to make, it wasn’t. The Drudge Report linked to the clip, and it immediately blew up. This was, as Rob Walker wrote for The Atlantic in an analysis of the event, a core principle of our new viral culture: “Humiliation should not be suppressed. It should be monetized.” Instead of being ashamed of this crappy television journalism, CNBC was able to make extra money from the millions of views it generated.

  The real reason the Santelli clip spread so quickly was a special part of toying with the valance of the web. Originally the clip spread as a joke, with the degree of amusement being determined by where the viewers fit on the political spectrum. But where some saw a joke, others saw a truth teller. An actual Chicago Tea Party was organized. Disaffected voters genuinely agreed with what he said. Santelli wasn’t having a meltdown, some thought; he was just as angry as they were. On the other end of the spectrum, not only were people not laughing, they were horribly offended. To them this was proof of CNBC’s political bias. Some were so serious that they endorsed a conspiracy theory (launched by a blog on Playboy.com, of all places) that alleged the meltdown was a deliberately planned hoax funded by conservative billionaires to energize the right wing.

  Regardless of how they interpreted Santelli’s rant, everyone’s reactions were so extreme that few of them were able see it for what it truly was: a mildly awkward news segment that should have been forgotten.

  Of all the political and financial narratives we needed in 2009, this was surely not it. Reasoned critiques of leveraged capitalism, solutions that required sacrifice—these were things that did not yield exciting blog posts or spread well online. But the Santelli clip did. CNBC fell ass first into the perfect storm of what spreads on the web—humiliation, conspiracy theories, anger, frustration, humor, passion, and possibly the interplay of several or all of these things together.

  As Chris Hedges, the philosopher and journalist, wrote, “In an age of images and entertainment, in an age of instant emotional gratification, we neither seek nor want honesty or reality. Reality is complicated. Reality is boring. We are incapable or unwilling to handle its confusion.”

  As a manipulator, I certainly encourage and fuel this age. So do the content creators. CNBC doesn’t care how they come off as long as they can sell ads against the traffic it brings. And the audience says they’re okay with it too—voting clearly with their clicks. We’re all feeding that monster.

  This may seem like nothing. It’s just people having fun, right? Sure, my deliberately provocative ads, once caught, quickly do disappear and awareness subsides—just like all viral web content. Roughly 96 percent of the seven thousand articles that made the Most E-mailed list in the New York Times study did so only once. In almost no cases did an article make the list, drop off, and then return. They had a brief, transitory existence and then disappeared. But though viral content may disappear, its consequences do not—be it a toxic political party or an addiction to cheap and easy attention.

  The omission of humanity from the popular slideshows of Detroit is not a malicious choice. There was no person like me behind the scenes hoping to mislead you. There was no censorship. In fact, there are thousands of the other, more realistic photos out there. Yet, all the same, the public is misinformed about a situation that we desperately need to solve. But heartbreaking sadness does not spread well. Through the selective mechanism of what spreads—and gets traffic and pageviews—we get suppression not by omission but by transmission.

  The web has only one currency, and you can use any word you want for it—valence, extremes, arousal, powerfulness, excitement—but it adds up to false perception. Which is great if you’re a publisher but not if you’re someone who cares about the people in Detroit. What thrives online is not the writing that reflects anything close to the reality in which you and I live. Nor does it allow for the kind of change that will create the world we wish to live in.

  It does, however, make it possible for me to do what I do. And people like me will keep doing it as long as that is true.

  * Another photo for a much more popular New York Times slideshow says it all. The picture is of the abandoned Michigan Central Station, and in the snow on the floor are dozens of crisscrossing footprints and a door. There are no people. “Don’t worry,” it seems to say. “There’s no reason to feel bad. Everybody left already. Keep gawking.”

  VII

  TACTIC #4

  HELP THEM TRICK THEIR READERS

  ARE LOADED-QUESTION HEADLINES POPULAR? YOU bet. As Brian Moylan, a Gawker writer, once bragged, the key is to “get the whole story into the headline but leave out just enough that people will want to click.”

  Nick Denton knows that being evasive and misleading is one of the best ways to get traffic and increase the bottom line. In a memo to his bloggers he gave specific instructions on how to best manipulate the reader for profit:

  When examining a claim, even a dubious claim, don’t dismiss with a skeptical headline before getting to your main argument. Because nobody will get to your main argument. You might as well not bother…. You set up a mystery—and explain it after the link. Some analysis shows a good question brings twice the response of an emphatic ex
clamation point.

  I have my own analysis: When you take away the question mark, it usually turns their headline into a lie. The reason bloggers like to use them is because it lets them get away with a false statement that no one can criticize. After the reader clicks, they soon discover that the answer to the “question” in their headline is obviously, “No, of course not.” But since it was posed as a question, the blogger wasn’t wrong—they were only asking. “Did Glenn Beck Rape and Murder a Young Girl in 1990?” Sure, I don’t know, whatever gets clicks.

  Bloggers tell themselves that they are just tricking the reader with the headline to get them to read their nuanced, fair-er articles. But that’s a lie. (I actually read the articles, and they’re rarely any better than the headline would suggest.) This lie is just one bloggers tell to feel better about themselves, and you can exploit it. So give them a headline, it’s what they want. Let them rationalize it privately however they need to.

  When I want Gawker or other blogs to write about my clients I intentionally exploit their ambivalence about deceiving people. If I am giving them an official comment on behalf of a client, I leave room for them to speculate by not fully addressing the issue. If I am creating the story as a fake tipster, I ask a lot of rhetorical questions: Could [some preposterous misreading of the situation] be what’s going on? Do you think that [juicy scandal] is what they’re hiding? And then I watch as the writers pose those very same questions to their readers in a click-friendly headline. The answer to my questions is obviously, “No, of course not,” but I play the skeptic about my own clients—even going so far as to say nasty things—so the bloggers will do it on the front page of their site.