Note: While reading a book whenever I come across something interesting, I highlight it on my Kindle. Later I turn those highlights into a blogpost. It is not a complete summary of the book. These are my notes which I intend to go back to later. Let’s start!

Arianna’s real superpower was her personal capacity for starting and maintaining relationships, for building a social network in real life. Her contact list contained nineteen thousand names. When Kenny arrived at her brightly lit home, with its faux-marble dining table and private office hidden behind a set of bookshelves, even he was impressed. The room was full of Hollywood royalty: David Geffen, Larry and Laurie David, producer Brian Grazer, writer Aaron Sorkin, the statesmanlike Norman Lear, Meg Ryan. Kenny spent most of the party trying to talk to Meg Ryan and listening to a litany of ideas that were, at best, kind of obvious: The internet was going to be important. Blogs were starting to matter. And perhaps most of all, the left needed its own version of Matt Drudge and his website, The Drudge Report.

Amid all the power players in that room, Arianna and Kenny turned out to be the ones who mattered most to what would come next. They shared an obsession with Drudge, a reclusive, gay Washington gadfly who had built his one-page site into the biggest and most important thing on the political internet, sprinkling just enough fresh gossip and witty headline writing on top of a quirky and partisan selection of links. He was the single most important force in traffic to news articles in America, driving some forty thousand views for a link in small Courier font on his site, and as many as a million—an unthinkable number, enough to crush most servers—for a big headline on the top of the page. That traffic meant that The Drudge Report wielded cultural power: Television producers looked at The Drudge Report to decide what to put on the air. New York Times reporters obsessed over its links. Ever since Drudge had broken the news, on January 17, 1998, that Newsweek was hanging on to a story about the president and an intern, Drudge had been the center of the world. And Drudge was a right-winger! He’d taken his talking points directly from the Bush campaign and had managed to propel unfounded claims that John Kerry exaggerated his Vietnam War record into the mainstream media. A leading political commentator of that moment, Mark Halperin, opined that Drudge was the Walter Cronkite of the aughts. Drudge was the man who set the agenda.

Kenny and Arianna wanted that kind of power, for themselves and for Democrats. Over the next few days, they resolved that they would build it together. Kenny was the moneyman, though he still wasn’t sure if this was a real business or just another political hobby. He found them office space down on what had become known as Silicon Alley, a couple of blocks of lower Broadway between Houston and Spring, where the lavish French bistro Balthazar was Nick Denton’s favorite place to bring bloggers he was trying to impress. Kenny put in $1 million of his AOL money. Huffington brought the social connections and media attention. People like the actor John Cusack and the historian Arthur Schlesinger Jr. agreed to blog for the new venture, even if they only hazily knew what a blog was.

From that day in 1995 forward, Andrew Breitbart had a front-row seat to the birth of the internet, as Drudge channeled the anger and resentment of their favorite talk radio host, Rush Limbaugh, into a riveting and simple page of links. There was right-wing politics and celebrity gossip, and other subtler strains that tickled America’s id: stories of loony liberals, terrifying Muslims, extreme weather, and weird science. Slowly but surely, the minor scandals that dripped out of newsrooms to the reclusive blogger became major ones. Andrew had been the silent force in the Clinton scandals, watching as the media raced to catch up with Matt’s (and his) obscure, ugly website.

If Harman couldn’t reduce his risk by knocking down the company’s valuation, he decided, he’d find other ways to lower his exposure in the deal. He insisted on “preferences”—terms that meant that The Huffington Post’s early investors would get paid after he did, and gave them an incentive to ensure that the company would sell for more than $300 million.And he told Kenny that he’d agree to the terms with one new condition: he’d invest only if Obama won the election. It made sense to Jonah, but Kenny was stunned. He’d never heard a condition like that. Harman wasn’t trying to make a political statement: his considered view was that Arianna and Kenny would have special access in an Obama presidency, that the news site would be aligned with the zeitgeist. It might have been unheard of to do, he knew, “but it was a particular moment in time.”On November 4, 2008, Obama won the election handily. All of The Huffington Post’s bets had come through at the same time. And the next December, Harman tagged along to the White House holiday party as Huffington’s plus-one. When they wound through the holiday decorations to the front of the photo line, where Barack and Michelle Obama towered over most of their guests, the president greeted the publisher warmly. “Arianna, I want you to know that I don’t agree with everything you say, but I read The Huffington Post every day,” he said.“That’s pretty cool,” Harman thought. “There’s really a special connection here.” Traffic might be the new currency, but, as Arianna had always known, proximity to power could also help pay the bills.

The money would pay for a new office on Spring Street, just a few blocks from Nick’s favorite haunt, Balthazar. Over lunch there soon thereafter, Jonah made clear to Nick that he’d been making a study of Gawker’s business. The Gawker Media sites, he told Nick, had gotten stuck in a “local maximum,” as writers responded to page view bonuses by trying to drive up clicks and “overserve the regulars,” as Nick later put it—at the expense of spreading their content across the wider internet and acquiring new readers. That bigger metric, the one that had supercharged The Huffington Post through the anonymous masses who surged in through Google links and viral stories, was “unique visitors.” Those people, too, could learn to click repeatedly, if Nick could only draw them in. Nick took Jonah’s advice and changed Gawker’s core metric to unique visitors—a metric that encouraged you to make content for everyone. Indeed, unique visitors would become the core metric of the whole viral internet. It was, Nick later realized, “a technical change, but fateful.”

With Google traffic gone, the editors scrambled to tap fully into social media—then composed of Twitter, a random feed of curiosities called StumbleUpon, Reddit, and most of all Facebook. Many of them had learned, from Jonah, an analytical approach: he taught them to think about why people would share something—not just because they enjoyed it, but because the act of sharing would bring them closer to their friends, or say something about themselves. Nobody understood this better than Jack Shepherd, the community manager, whom Jonah had hired away from the animal rights group PETA. Shepherd specialized in simple, direct images of—for instance—a person rescuing a baby goat from drowning. These were categorized on Reddit, where BuzzFeed editors frequently turned for their memes, as “restoring your faith in humanity.” Shepherd liked to brag that for a time, the second-most successful post in the history of BuzzFeed was a simple, inspiring list of images that would, in fact, make you feel better about your fellow human beings. Its success showed the way in which social media brought out people’s best impulses, the ones they were proud to share with their friends; it gave hope that BuzzFeed and its ilk would pull the internet out of secrecy of a search bar in which middle-aged men were typing “Rihanna topless,” and into the light of Facebook, where you were more likely to share the posts that made you look like a good person, the kind who would rescue a goat yourself. (Of course, that list had been the second-most popular; Jack often glossed over the most popular post, a list of “fails”—people falling off escalators and walking into glass doors.)

Upworthy’s technical proficiency at maximizing traffic returns was also a vulnerability, and it left Upworthy open to attack from inside Facebook. Jonah had begun to make a habit of cultivating the mid-senior-level Facebook employees who ran its key product, News Feed. He invited them by BuzzFeed’s familiar-feeling offices when they were in New York, and stopped by to say hi when he was in the Bay Area. Over coffee in San Francisco, and in spontaneous direct messages on Twitter, Jonah could offer them the one thing Facebook didn’t have: insight into how traffic was moving around its rival networks, Twitter and Pinterest. What’s more, unlike most in media, he seemed to speak their language, without clumsy simplification from the pompous jargon of journalism. “It was easier to talk to him without having to translate than it was to talk to most media executives,” according to a top Facebook executive. They knew Jonah was pushing them, trying to persuade them to shape their service in a way that would help BuzzFeed—but it was nice to talk to someone who thought this new medium could make the world better, who saw you as an ally, not an alien.Jonah had invested in his relationships with Facebook, and they were paying off, because Facebook was now unquestionably the world’s most important source of traffic. Websites across the industry showed a massive spike beginning in August of 2013, as Facebook referrals spiked 69 percent by October. “Can Mark Zuckerberg save the publishing industry?” BuzzFeed’s Charlie Warzel wondered. Upworthy seemed unstoppable, rising to a peak of eighty-seven million monthly views in November of 2013. That month a top News Feed engineer stopped by BuzzFeed, and after he left, Jonah wrote to him about what he saw as a persistent problem in Facebook’s feed, the one Upworthy was particularly famous for exploiting. “Right now there is an incentive for publishers to write incomplete headlines to get people to click, aka ‘click bait’ aka ‘the curiosity gap,’ ” Jonah wrote. “But those headlines fill News Feed with text devoid of information, unlike descriptive headlines that provide value even to users who don’t click them. Although this gets more people to the story and many of those people might share the story, the 90%+ who never click the story get no value at all.”Jonah’s belief that these headlines were essentially spam led him to instruct me—to the disappointment of some of my traffic-hungry writers—to ban them from BuzzFeed. As he wrote in the email, while other sites might write a headline like “You Won’t Believe What Disneyland Forced All Employees to Do in 1965,” BuzzFeed’s headline was “Disneyland’s 1965 Employee Handbook Was Just as Strict as You’d Imagine.”“If you don’t click the story you still get the basic point that Disney was strict and conservative in the 60s,” Jonah pointed out. “Is there a good way to measure this discrepancy in value? It is something a human editor can do easily, especially one focused on building long term trust with a reader, but feels trickier to automate.”Jonah forwarded me the exchange, adding, “It is really fun collaborating with Facebook’s team on how News Feed should work.” When Facebook figured out how to purge clickbait headlines, Upworthy’s traffic took the hit.

Eli occupied a strange position. He had built his company on top of Facebook’s algorithms, but he was also a Facebook critic. His own startup had proved out his dark theory about the filter bubble. Facebook kept trying to refute his argument: their own research, they said, showed that the information you saw on the site was dictated by whom you chose to be friends with—not by Facebook’s algorithms.But Facebook’s research contained a curious detail, one of the company’s own analysts confided in Eli: relatively small right-wing websites seemed to be getting far more “engagement” by Facebook’s metrics than huge liberal ones, like Upworthy, or The Huffington Post, or BuzzFeed. If the progressives were getting all the traffic, Eli asked, what did this “engagement” figure mean? The researcher confessed to him that the company was puzzled by it, too, by the people sharing and resharing posts from clunky old conspiracy websites like WorldNetDaily and obscure lists of links at “There’s this pretty much epistemically closed group of people who are super-sharers,” he was told. “It’s this alternate universe on Facebook that’s behaving differently than anyone else.”It wasn’t the only sign of a cultural turn. On discussion boards devoted to video games, a newly overt anti-feminist sentiment was brewing. Some of the figures of the early internet were being drawn into that circle. Tracie Egan first took her old friend Gavin McInnes’s use of the N-word as ironic rather than racist, but she fell out with him when he began using the word full-throatedly in 2012, she later recalled. And the new clusters of sharing that the researcher had described to Eli were among the first signals that Facebook wasn’t quite what the liberal utopians of Silicon Valley and SoHo thought it was, and that a different kind of politics could displace the optimism of Morales and Obama. Facebook’s audience had become older, now reaching far beyond the young college-educated elites. And Facebook, it turned out, didn’t much like Upworthy, with its irresistible headlines and compelling videos whose function was, mostly, to trick you into leaving Facebook. Upworthy depended on Facebook, but to the platform, it was little more than a bug. A series of decisions—killing those headlines, and encouraging users to watch videos directly inside Facebook instead—would begin to choke off traffic to Eli’s project. Those curious clusters of intense engagement, of older, whiter, more right-wing Americans spending more and more time liking and sharing the same articles, and clicking on advertisements as they spent hours on the site—that was what Facebook was going for, that was the future Mark Zuckerberg chose. Andrew Breitbart was dead. But the future he’d imagined was only starting to come to life.

BuzzFeed’s editors had learned through trial and error that social media was organized in large part around identities. Write compellingly about what it was like to be born in the 1990s, or to be an Iranian from New Jersey, or a Catholic girl, or to grow up with East Asian parents, and thousands of people would share the post with the magic words “this is me.” Their friends would click for a glimpse of insight. The traffic was guaranteed, and the posts extended BuzzFeed’s tendrils into new territories.To reach a broader audience, we needed people who could write about varied identities. That pushed BuzzFeed, whose small staff was nearly all white in its early days, to become among the most ethnically diverse of the new media companies. But we were still missing a big piece of American identity. We didn’t have any proud young conservatives, people who could write about what it was like to grow up with guns, say, or to appreciate how the Bush family respected veterans. Benny represented, to me, an untapped new well of traffic, a new identity to plumb. And so I didn’t look much beyond that NRA post, which took the BuzzFeed formula—a list of fun, emotionally resonant images—to gun culture. There was no reason that Tea Partiers couldn’t see themselves in BuzzFeed, I reasoned, and share elements of their culture—guns, cars, Bibles—just as our progressive audience was doing. I realize now that I allowed my eyes to skate over some of Benny’s other hits: for instance, his recurring focus on the obscure New Black Panther Party, and race-baiting headlines like “Don’t Miss the Connection: Obama ‘Delivered’ to Office by Black Panthers, Holder ‘Owes Them Some Favors.’ ”

In those early days at BuzzFeed, we didn’t pay much attention to job titles (or pay scales, or much in the way of organizational structure), and most writers and reporters didn’t care what they were called. Benny, though, chose a very specific title for himself: “viral politics editor.” He positioned himself at the intersection of the kind of analytical approach to content that Jonah had pioneered with experiments like LilyBoo, and my own fixation with political journalism. And while Benny’s roots were on the right, his core obsession wasn’t that different from anyone else’s at BuzzFeed: he wanted traffic. His new colleagues were skeptical of him—Media Matters had released a dossier on Benny’s various right-wing views when I hired him—and when he came to New York from his Washington home, Benny was often the only one besides me wearing a blazer. But he had a gift for traffic, and at first he seemed drawn mostly to the pageantry of power: he delighted the internet with visual accounts of interns sprinting to deliver Supreme Court verdicts, or of the Dunkin’ Donuts in the grim Capitol basement. He won his colleagues over quickly enough, when they went out after work to a gay bar on West Twenty-Fourth Street, XES Lounge, that had the cheapest drinks in the neighborhood, and he bought a couple of rounds of shots. He’d done his homework, too, and praised the work of each new person he met.

Later, I’d recognize in Benny something we’d come to associate with the president who was elected three years later. The thirst for attention, the willingness to say absolutely anything to go viral, and the attraction to uniforms and rituals of power. Benny kicked around in the media industry awhile longer, getting fired for allegedly being lazy or erratic or for taking credit for others’ work. Meanwhile, the world was turning his way.

The BuzzFeed post was so simple: a quiz asking “What State Do You Actually Belong In?” It was the latest in a rush of quizzes that were, for reasons not totally clear at the time, getting enormous amounts of traffic from Facebook. The format was as old as women’s magazines like the ones Anna Holmes used to work for—“What’s Your Secret Sexual Personality?” And the substance had nothing to do with the increasingly ambitious journalism we were starting to pull off. But the technology to do it simply and shareably on the web was, in that moment, new, and ours. The questions were not, let’s say, scientific: “Which animal do you feel best represents you?” it asked. “What’s your party anthem?”We gathered, astounded, to watch the numbers build on Google Analytics, the giant company’s dominant entry into the field that Jonah had helped invent, and that Chartbeat had dominated before Google got into the game. Traffic to the states quiz built to around 200,000 people on the site at the same time—about twice as many people as we’d see across all of BuzzFeed most days. Our Amazon-hosted servers held.We knew the quiz was working. We weren’t sure why. Then Nguyen, who ran BuzzFeed’s analytics, noticed something: there was a bug in the quiz. If your answers to the random, silly questions produced a tie between two results, the quiz defaulted to the one lower down in the alphabet. This meant a lot of people got Wyoming. More people, we calculated, than the 582,000 people who actually live in Wyoming. Many of them took to Facebook to complain about it. “Is that still a state?” one reader asked. “I hate the freezing weather. That’s why I am in California. ,” said another. “How the hell did I get Wyoming?” asked a third. And so on.All the results were even less precise than our usual efforts to vaguely associate your pizza preference with your personality type. We couldn’t have cared less if the quiz actually matched you to your dream state—it was something to do while you were bored at work, and then it turned into something to argue about online. But if we saw good-natured complaints on our Facebook page, Facebook saw something else: engagement. It didn’t really matter what people were saying. What mattered was that they were talking at all, spending more time on the platform, feeling and reacting.

The Facebook crew knew Jonah was working them, for information and to shape their thinking in a way that would match his vision for the internet. But they didn’t mind it. They were tired of translating their own point of view—about audiences and content and measuring what people really liked—to people who spoke the old media’s language, of viewers and readers, and the journalistic and aesthetic standards that determined what they ought to like. What’s more, Jonah had also proven willing to warn Facebook when its algorithm was inadvertently promoting spammy formats that wouldn’t wear well with its users, to flag bugs rather than taking advantage of them, as he had with those “curiosity gap” headlines on which Upworthy had lived and died. Jonah warned Facebook of short-term tricks media companies were using to game its system, and he warned his employees not to play those games. And in return, he got insight into how Facebook was thinking, and why all those people who were definitely, certainly not moving to Wyoming had added up to the most successful post in BuzzFeed’s history so far.The particular piece of insight he brought back from a conversation with a Facebook staffer in 2014 was a simple piece of math: 1 comment = 4 likes. The company’s algorithm had shifted to prioritize “engagement,” rather than giving a simple thumbs-up. We’d entered a new world, one that Facebook and its critics alike would sometimes conceal behind mystifying technical language, but which boiled down to two things: metrics based on how much people were sharing, clicking, and talking about a piece of content; and an algorithm that promoted the pieces that drew that engagement—not just to engaged individuals’ friends, but to everyone on the vast network. The details would change over time, but the formula wouldn’t—and a new global wave of engaging right-wing politicians would be ready for it.

Donald Trump was made for Facebook. So was another type of content: attacks on Hillary Clinton. Among the first to discover that fact were a handful of teenagers in Veles, Macedonia, a small city on the Vardar River, who wanted some spare cash. Their discovery was of a sort of arbitrage: if they could direct Americans, rather than Macedonians, to their websites, they’d be getting paid in dollars, not denars. And even as American publishers had begun to suspect that traffic wasn’t quite like oil—a limited commodity that would hold its value—and we were beginning to grumble about trading print dollars for digital pennies, well, digital pennies were still pretty good in Veles. The teenagers created websites like and told people on Facebook exactly what they wanted to hear: for instance, that Hillary Clinton would soon be indicted. “Your Prayers Have Been Answered,” declared one headline. Another falsely reported that the pope had endorsed Donald Trump.

A few weeks before the election, I got a lesson on this phenomenon in person from Steve Bannon, who had taken over when Andrew died, then left it to go chair Donald Trump’s campaign. He’d been interested in Jonah since his first glimpse at the business of The Huffington Post, so I’d been invited to Trump Tower to exchange notes with him. I walked past the golden escalator and rode an elevator up to an empty conference room to find the chairman alone, and not in any particular rush. Bannon exuded utter confidence, but it didn’t feel like a winning campaign. He didn’t seem to have much to do. And he told me he was puzzled by BuzzFeed. Breitbart hadn’t just chosen Trump, he told me, based on the candidate’s political views. Bannon and his crew had seen the energy Trump carried, the engagement he’d driven, and attached themselves to it. BuzzFeed, in Bannon’s view, had failed to recognize that Bernie Sanders could generate the same energy, the same engagement. Why hadn’t we gone all in for Bernie? he asked me.Jonah had sometimes asked the same thing. He, too, saw that the energy was on the militant left, and that our staff’s sympathies—and his own—mostly leaned the same way. But our journalistic scruples, the impulse toward fairness and away from propaganda, sometimes handcuffed our drive for traffic. Our news operation had gone from being scorned and yoked to the social web to being well regarded among journalists, and even winning the occasional prize—but we were spending more and more money on journalism to maintain the same level of traffic, and though he and I didn’t talk about it much, Jonah and I both knew that our existing approach to news depended on the rest of the business remaining strong. But Jonah had bet on me and on that style of news, and so Jonah let me win the occasional skirmishes with our publisher, Dao Nguyen, when he eyed the traffic that a breathless Bernie Sanders fan post might get. I told Bannon that we came from a different journalistic tradition, and we valued it. That answer didn’t satisfy any of us much.