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

by Ryan Holiday

  XV

  CUTE BUT EVIL

  ONLINE ENTERTAINMENT TACTICS THAT DRUG YOU AND ME

  YOU SIT DOWN TO YOUR COMPUTER TO WORK. FIVE minutes later you’re on your fifth YouTube video of talking babies. What happened? Do you just not have any self-control? Sorry, but self-control has got nothing to do with it. Not when the clip was deliberately made more attractive by subliminally embedded images guaranteed to catch your attention. Not when the length of the video was calibrated to be precisely as long as average viewers are statistically most likely to watch.

  Would you also be surprised to hear that the content of the video was designed around popular search terms? And that the title went through multiple iterations to see which got the most clicks? And what if the video you watch after this one (and the one after that and after that) had been recommended and optimized by YouTube with the deliberate intention of making online video take up as much time in your life as television does?1

  No wonder you can’t get any work done. They won’t let you.

  The key, as megawatt liberal blogger Matt Yglesias advised when interviewed for the book Making It in the Political Blogosphere, is to keep readers addicted: “The idea is to discourage people from drifting away. If you give them a break, they might find that there’s something else that’s just as good, and they might go away.”

  We once naively believed that blogs would be a boon to democracy. Unlike TV, the web wasn’t about passive consumption. Blogs were about engagement and citizen activism. Blogs looked like they would free us from a crummy media world of bias, conflict, manipulation, and sensationalism. But as James Fennimore Cooper presciently observed in the nineteenth century, “If newspapers are useful in overthrowing tyrants, it is only to establish a tyranny of their own.”

  Tyranny is an understatement for the media today. Those between the ages of eight and eighteen are online roughly eight hours a day, a figure that does not include texting or television. America spends more than fifty billion minutes a day on Facebook, and nearly a quarter of all Internet browsing time is spent on social media sites and blogs. In a given month, blogs stream something like 150 million video streams to their users. So of course there is mass submission and apathy—everyone is distracted, deliberately so.2

  The idea that the web is empowering is just a bunch of rattling, chattering talk. Everything you consume online has been “optimized” to make you dependent on it. Content is engineered to be clicked, glanced at, or found—like a trap designed to bait, distract, and capture you. Blogs are out to game you—to steal your time from you and sell it to advertisers—and they do this every day.

  THE ART OF THUMBNAIL CHEATING

  You see a link to a video in a YouTube search that makes it look like a hot girl is in it, so you click. You watch, but she’s nowhere to be found. Welcome to the art of “thumbnail cheating.” It’s a common tactic YouTube publishers use to make their videos more tantalizing than the competition.

  The most common play is to use a girl, preferably one who looks like she might get naked, but it can be anything from a kitten to a photo of someone famous. Anything to give the clip an edge. Some of the biggest accounts on YouTube were built this way. The technique can drive thousands or even tens of thousands of views to a video, helping it chart on most viewed lists and allowing it to spread and be recommended.

  Online video publishers do this with YouTube’s consent. Originally, YouTube chose a video thumbnail from the ½, ¼, or ¾ points of the video. So smart manipulators simply inserted a single frame of a sexy image at exactly one of those points in order to draw clicks. Members of the YouTube Partner Program—the people who get paid for their contributions to YouTube through ad revenue and make millions for the company—are allowed to use any image they choose as their thumbnail, even images that don’t ever appear in the video. Sure, YouTube asks that the image be “representative” but if they were actually serious about quashing profitable trickery, why allow the practice at all?

  GENETICALLY MODIFIED ENTERTAINMENT

  LOLcats, the cute captioned kitty photos, are a viral mainstay that started as good-time fun but are much more than that now. It’s not enough that some may make you chuckle while others may not. A hit or a miss is a risk that must be avoided.

  In May 2011, the Cheezburger Network—now also the powerhouse purveyor of fail photos, funny infographics, and daily links, with nearly a half-billion pageviews a month—hired a prominent data scientist. His job: to build a team to monitor every pageview and metric the sites get in order to shape the content around that information. That is, in his words, to engineer “more smiles for people per day.” A media empire paid by the smile can’t afford anything less.

  I mean no disrespect. After all, I sold an Internet meme site I owned called FailDogs.com to the Cheezburger network. I knew I was never going to be as good as they were. I was just one person, and I couldn’t turn the fifteen minutes of fame from the site into a business. But Cheezburger could, by rendering users powerless to resist the urge to click. And they could do it with an irresistible veneer of cuteness masking their tactics.

  Entire companies are now built on this model, exploiting the intersection between entertainment, impulse, and the profit margins of low-quality content. What they produce is not so much information but genetically modified information—pumped with steroids and hormones.

  Demand Media, owner of eHow, Lance Armstrong’s Livestrong.com, Cracked.com, Answerbag.com, and others, specializes in this type of algorithmically created media. Relying heavily on computer algorithms and massive data dumps, they craft online perfection in the form of low-cost, click-heavy content that advertisers love. Like successive sieves, each refines the contents of the one that came before it; Demand’s automated editing systems pump out up to thirty thousand video clips and articles about trivial topics like baking cookies or “best of” lists. It generates millions of pageviews a day, and all of it sucks.

  Their process is simple. First, Demand’s algorithm trolls the web for lucrative search terms. It dreams up a piece of media, such as a video tutorial or a brief article, that combines as many popular terms as possible and estimates a lifetime value (LTV) of its financial worth. A second algorithm examines this concept again, creating options for the most search-friendly and provocative title. These options are fed to a human editor trained in the same art, who selects the best one. Then another editor reviews the previous editor’s choice and optimizes it further, before settling on the final pitch for what should be created.

  It is here, after being processed through secret computer algorithms and surgically modified by data analysts instead of editors, that the product is finally ready for writers. These writers are paid to follow the exacting prescriptions of more data-driven instructions. By the time the content is ready to be published, advertisements will have already been sold against it. These advertisers are Demand’s real audience.3

  When these content rules are not explicitly mandated by data specialists and analysts, they are implicit; bloggers know to default to what will spread and please the advertisers. People taught the logic of machines are Demand’s final sieve. As one Demand Media editor e-mailed to a new contributor whose first article was rejected for not following their surefire format for going viral: “The mistakes you’ve made indicate you’re new to Demand. This will become second nature as you learn the formats and the site requirements.”4 It’s a second nature known well by YouTubers, LOL makers, podcasters, bloggers, and tweeters.

  DRUGGED AND DELUSIONAL: THE RESULT

  I remember seeing Jeff Jarvis, the blogger best known for his condescending (and unsolicited) advice to the newspaper industry, at a tech conference once. He sat down next to me, ostensibly to watch and listen to the talk. Not once did he look up from his laptop. He tapped away the entire time, first on Twitter, then on Facebook, then moderating comments on his blog, and on and on, completely oblivious to the world. It struck me then that whatever I decided to do with the res
t of my life, I did not want to end up like him. Because at the end of the talk, Jarvis got up and spoke during the panel’s Q&A, addressing the speakers as well as the audience. In the world of the web, why should not paying attention preclude you from getting your say?

  That’s what web culture does to you. Psychologists call this the “narcotizing dysfunction,” when people come to mistake the busyness of the media with real knowledge, and confuse spending time consuming that with doing something. In 1948, long before the louder, faster, and busier world of Twitter and social media, Paul Lazarsfeld and Robert Merton wrote:

  The interested and informed citizen can congratulate himself on his lofty state of interest and information and neglect to see that he has abstained from decision and action. In short, he takes his secondary contact with the world of political reality, his reading and listening and thinking, as a vicarious performance…. He is concerned. He is informed. And he has all sorts of ideas as to what should be done. But, after he has gotten through his dinner and after he has listened to his favored radio programs and after he has read his second newspaper of the day, it is really time for bed.5

  This is the exact reaction that web content is designed to produce. To keep you so caught up and consumed with the bubble that you don’t even realize you’re in one. The more time kids spend online, studies show, the worse their grades are. According to Nielson, active social networkers are 26 percent more likely to give their opinion on politics and current events off-line, even though they are exactly the people whose opinions should matter the least.

  “Talkativeness is afraid of the silence which reveals its emptiness,” Kierkegaard once said. Now you know why sharing, commenting, clicking, and participating are pushed so strongly by blogs and entertainment sites. They don’t want silence. No wonder blogs auto refresh with new material every thirty seconds. Of course they want to send updates to your mobile phone and include you on e-mail alerts. If the users stops for even a second, they may see what is really going on. And then the business model would fall apart.

  XVI

  THE LINK ECONOMY

  THE LEVERAGED ILLUSION OF SOURCING

  IN 2010, AFTER MANY YEARS OF SUCCESSFULLY TRADING bogus stories up the chain, I was in the ironic position of desperately trying to stop it from happening—at the highest of levels: CNN. It was more than just karma. When you feed the monster as I had, it will eventually come back and attack you.

  This was the situation: A disgruntled store manager sent e-mails to Gawker “exposing” what he or she claimed were discriminatory hiring practices at American Apparel. Why Gawker? Because he or she knew that Gawker loved to write about the company—snarky blog coverage I had encouraged both directly and covertly in the past. The manager alleged that the company refused to hire “ugly people,” and supposedly enforced this policy via photographs sent to corporate headquarters. Gawker ate it up.

  The manager’s anonymous e-mails, along with a handful of “leaked documents” about American Apparel’s dress code, were published on the site as proof that the accusations were true. There was only one problem. Not only were the practices not discriminatory—legally or morally—but they were not even new. The same dress code had been written about nearly a year earlier by other blogs.

  More important, asking for a photograph of a retail applicants’ personal style was far from invasive surveillance. American Apparel isn’t Panopticon. The company was simply looking to make sure that managers hired the kind of real people who shopped in our stores—expressly reducing the pressure for cosmetic alterations like breast augmentation, heavy makeup, tattoos, piercing, plucking, hair dying, and straightening that most retailers actually do select for. We were specifically trying to reduce discrimination. Not that the leaked documents were some smoking gun, anyway. What Gawker had was a hodge-podge of unsanctioned and unverified notes from low-level employees, style advice from the creative department, and little else.

  The controversy was a farce. The only source was the anonymously complaining ex-employee, and even then their claims were significantly exaggerated by the sites that wrote about them. I watched as this initial report from Gawker spread from sites across the news spectrum, getting bigger and more outrageous with each new mention. Fashion blogs turned the accusations into proven fact; the anonymous words of the ex-employee became “officially stated policy” on others. Stock blogs “analyzed” the effect the policy would have on the stock price. Other news blogs rolled up other allegations to take the story to new levels—like remarks supposedly overheard by chattering retail employees that they represented as company statements.

  It came to a head when a reporter at CNN—the top of the news chain—contacted me, because they had watched the story develop and wanted to report on it.

  This was our e-mail exchange (edited only for formatting and length):

  To: Ryan Holiday

  From: CNN

  CNN is covering the story about you and your possible hiring practices reported by gawker.com. Would you be able to answer the accusations on CNN this Saturday evening either in the 5pm or 7pm hour?

  The key bullshit word here is “possible” placed right before “hiring practices.” Obviously the reporter believed these were the actual hiring practices or they wouldn’t be doing a story. But since CNN couldn’t report on just rumors, they wanted to make it a story by getting me to deny it. I knew this was an attempt to pretend there were two sides to the issue. But there weren’t two sides; there was simply the truth and an untruth.

  To appeal to the reporter’s sanity and expose how this story had been traded up the chain, I responded with the following e-mail, after passing along the company’s official statement:

  To: CNN

  From: Ryan Holiday

  Hopefully you can tell from our statement that the Gawker report is probably misconstrued at best, possibly inaccurate in other areas. It’s important to point out that the verification and anonymous sourcing politics for blogs and the one that you surely have at CNN are very different and can’t be conflated.

  It’s unfair and inaccurate to hold this up as being something the company engaged in primarily based on the fact that another less rigorous outlet mentioned it first. What we attempted to say in the statement was that as a company who has always challenged beauty and diversity norms in the fashion industry—not quietly but as the central part of our creativity—accusations like that are not only unfounded but are contrary to what we’re committed to. What I was attempting to convey in my original emails is that in the past outlets have used the vehicle of “reporting on what _____ is reporting” to include information they likely wouldn’t have included through their own editorial standards. Hopefully CNN does not do that.

  After a long pause, the reply:

  To: Ryan Holiday

  From: CNN

  Subject: CNN no longer doing Gawker story segment

  After a lot of consideration we decided to no longer do the segment.

  Though I narrowly dodged a bullet with CNN, it was during this incident that I began to understand the web’s sourcing problems from a new perspective. A dubious accusation on a gossip blog nearly became a frighteningly nongossip story from the “most trusted name in news.” There had been no overt manipulation, yet something completely untrue had spread from one site to another as though some invisible hand had guided it along. Thankfully, it did not make it to air on CNN, but it could have had I not stepped in.

  Henry Blodget, in a revealing onstage interview with reporter Andrew Sorkin, explained the increasingly common cycle like this: “There are stories that will appear on Gawker Media—huge conversations in the blogosphere—everything else. It’s passed all over. Everyone knows about it. Everybody’s clicking on it. Then, finally, an approved source speaks to the New York Times or somebody else, and the New York Times will suddenly say, ‘Okay now we can report that.’”

  This is exactly what happened with the near CNN and Gawker debacle. A story that originated on G
awker as a controversy was the center of an enormous amount of buzz online. It then grew and grew as it spread from site to site, until the buzz was noticed by CNN, which tried to get me to discuss the story with them on air. CNN, of course, would never have been able to break the story themselves, nor would they have been interested in bothering with something so small as a manager’s anonymous e-mail. But if someone else made it a hot topic first, they were happy to do their own piece on it. It’s the same tactic I abuse when I turn nothing into something. Get a small blog to pick a story up and pass it upward to bigger and more credible outlets, which simply link to the previous report and don’t bother to verify it.

  Both the blogs and the mainstream media are shirking their duty—and that makes them ripe for exploitation (or in the case of American Apparel and CNN, a missile that can strike your company at any time). And yet most of the social media elite want this for our future.

  THE DELEGATION OF TRUST

  This cycle has roots in two journalistic habits—one from the new media world and one from the old. When combined, they become a major danger.

  Reporters can hardly be everywhere at once. For most of recent history, media outlets all used the same self-imposed editorial guidelines, so relying on one another’s work was natural. When a fact appeared in the Chicago Tribune, it is pretty safe for the San Francisco Chronicle to repeat that same fact, since both publications have high verification standards.