DAVE MORIN WAS a Montana kid who grew up on computers. He’d paid his way through the University of Colorado by running a web-development firm out of his dorm room. After graduating in 2003, he joined his dream company, Apple, working for its higher-education marketing team. His job was getting college kids to use Apple tools, and he was put in charge of its campus-representative program. At the time, there were about 100 reps across the United States, mostly geeks, and their task was providing technical support to their peers. Morin shifted the program to evangelism, expanding to 900 students who pitched Apple to classmates. Morin believed in communities, and he always urged his reps to join social networks—Friendster, LinkedIn, even AIM. One day in early 2005 the rep from Harvard called him. You have to see this thing called Thefacebook.
Morin still had his .edu address from Boulder and he signed on. He was stunned. One of his complaints with AIM, which was the de facto communications system of college students then, was that you couldn’t find anyone or know who they were because everyone used oblique screen names. Facebook used your real name—and it also provided your AIM handle in the profile. He was also dazzled at how privacy was elegantly baked into the network structure—you could browse or message anyone on your campus, but not in other schools. Game over, he thought. This is genius.
He immediately tried to get hold of whoever was running this thing, and soon found himself in the tiny Palo Alto office whose walls looked like they had been tagged by talented and horny vandals. The leader, Zuckerberg, was clearly super intelligent, but he hardly uttered a word. Morin bonded more with Moskovitz and Parker.
At that point, the only other major brand working with Facebook was Paramount Pictures, which ran a big promotion on Facebook for its SpongeBob SquarePants movie. Morin wasn’t interested in buying ads. He wanted to start a Facebook group to promote Apple—someplace where people could learn about the products, share videos and other content, and exchange tips about using Macs. Apple would lure them there with giveaways of iPods and iTunes cards. They cut a deal where Apple paid Facebook $25,000 a month. The total contract might have been a million dollars. Parker would boast about the contract when negotiating the Accel deal.
By then Morin had broken through Zuckerberg’s ice-wall demeanor. Morin and Zuckerberg and Moskovitz would have endless conversations about graph theory, identity theory, signal theory. The latter dealt with how humans signal their identity by things like status indicators. Morin came to understand that Facebook was the ultimate status indicator, as well as a lubricant for a new social order. It was a workshop for the way people would live with one another in the future.
Zuckerberg and Moskovitz urged Morin to join Facebook, but it was hard to leave Apple’s beautiful headquarters for a start-up crazytown in downtown Palo Alto. Once Moskovitz and Ezra Callahan visited Morin at Apple’s sprawling Infinite Loop campus. “This place is pretty nice,” they told him. “But one day we’ll be bigger.”
Really? Morin thought. Come on!
Morin tried to get his bosses at Apple excited about Facebook. His dream was for Apple to make a social operating system. Instead of organizing your system around files, why not around people? Maybe Apple could buy Facebook, as the basis of this new system. The matter came before CEO Steve Jobs. No go. Jobs was open to buying companies, but why join forces with a college-only site of a few million people when MySpace had fifty million?
Morin kept talking to Facebook. One day in the fall of 2006, Moskovitz visited him at Cupertino again. Wouldn’t a social operating system be amazing? Morin asked him. Moskovitz point-blank stared at him. This was what they were talking about at Facebook all the time! You need to come to Facebook now and do this, he told Morin.
Steve Jobs had just given his famous graduation speech at Stanford where he told students to approach each day with the realization that death could come anytime. That gave Morin the courage to cut the cord with the company he’d always wanted to work with and join the company he thought was poised to make an even bigger splash. He approached Jobs at an employee event and told him he looked in the mirror that morning and realized he had to go to the start-up he’d been gushing about. Jobs asked only one question. Are they offering you a good stock package?
Yes, they were. Dave Morin is now worth at least a hundred million dollars.
That weekend the newly hired Morin was at Facebook’s office talking to Zuckerberg. It was late at night, in the corner room that Zuckerberg often used for one-on-ones. It was all white—white table, white walls, white Eames chairs, and pretty much all the wall space was covered with whiteboards. People called it the Cloud Room. But sometimes it could feel more like an interrogation room.
Zuckerberg told Morin that while Apple was an innovation company, Facebook was a revolution company. Morin felt a jolt of energy. For the first time, he felt he understood Zuckerberg and Facebook. Facebook creates revolutions.
And Morin would be part of it, creating the platform that would catapult Facebook into the ranks of the very top tech companies. It was already in the works.
Platform.
THE FIRST ADHERENT of the platform was an engineer named Dave Fetterman, who had arrived at Facebook in January 2006. Hailing from York, Pennsylvania, he had graduated Harvard the year before Thefacebook appeared, and taken a job at Microsoft. He was among a group of ace engineers in their mid-twenties who left Seattle for Facebook that winter and became known as the Microsoft Five. (Andrew Bosworth was also in the group.) The newcomers took a house together that they dubbed Facebook Frat. The first task Fetterman performed was adding a few more options to the “relationship status” field on profiles. This was the moment when Facebook added “it’s complicated” to the choices. (A phrase destined to launch a thousand Facebook headlines in later years.)
As he checked off task after task at the growing company, Fetterman couldn’t stop thinking of a casual question that Moskovitz had tossed out at his job interview: What would a Facebook development platform look like? A “development platform” would mean a technological gateway for other software developers to create programs that used the data from Facebook for social applications. The first step would be to create an API, or application programming interface, which was kind of a software socket that people could plug their programs into, allowing access to the data on the platform.
Fetterman asked Moskovitz if he could write that API. Moskovitz said no. The next week he asked again, with the same answer. Finally, Fetterman decided to just do it. He built the gateway, and coded up a prototype application, something software developers might create to make use of the API. It was called Owen Van Natta’s Balloon Store. Using the API, the app knew the birthdays of Van Natta’s friends, which it could access from Facebook. “It was the ugliest piece of HTML you’d ever seen,” he says.
Fetterman demoed this to his colleagues. Wouldn’t it be great, he asked them, if, say, you could go to Amazon and find out what your friends were reading? Or you could go to any site and find out what your friends were up to there? It would be like Facebook was everywhere!
It would also mean that the birthday app would now know the birthdays of Van Natta’s friends though they did not give permission or even know about the transfer of information. And that if someone used that Amazon app, the world’s biggest bookstore would know the reading habits of that person’s friends without their knowing it.
It was a problem that Facebook would grapple with for years.
Fetterman’s idea came before what was then the Facebook brain trust. The reaction was almost unanimous—Why would we give away our network? As Fetterman recalls, only one person thought it worth pursuing.
“I think we should look into this,” said Mark Zuckerberg.
That summer, Facebook released Fetterman’s API. It flopped. “We said, Hey come one, come all, use the Facebook platform to build interesting things,” says Fetterman. “But no one noticed.”
It turns out that simply releasing an API wasn’t enough. For one thing, there had to be a way for Facebook users to know that there was some other social application using the API, and that their friends were using that app as well. It was a problem of distribution.
As it happened, Facebook itself was in the process of creating one of the most effective ways of distributing data about everybody’s friends—the News Feed. Why not employ it to lure users to apps running on the new platform?
Facebook and Zuckerberg were eager to keep pursuing the concept. That had been their reason for hiring Morin: to spearhead developer relations. Facebook’s chief technical officer, Adam D’Angelo (finally full-time now that he graduated Caltech), took charge of the platform-engineering team, with Fetterman as the tech lead. After endless whiteboard sessions, Fetterman’s original idea of just an API evolved to a much broader undertaking, where the apps would not be on someone else’s website but Facebook’s, living on pages that were called canvases. Users would learn about them through the News Feed.
“We said here’s a canvas inside the trusted blue-and-white borders where you could build whatever your dreams dictate,” says Fetterman.
The difference between plan A (Fetterman’s original API) and the new plan B was that the latter positioned Facebook as not just a platform but an operating system. This was the pinnacle of Silicon Valley’s pyramid of value. If you owned the OS, you had your own little monopoly. The most successful OS of the previous era was Microsoft’s Windows, which a judge had determined was in fact a big monopoly. While many Silicon Valley leaders still viewed Microsoft as the industry’s Darth Vader, Zuckerberg admired Bill Gates’s company. The Windows system was unbeatable because a huge majority of PC users had computers that ran it. To reach those customers, software programmers had to write their software in Windows. Zuckerberg came to imagine Facebook as the social equivalent of that. Just as Microsoft owned the desktop world, Facebook would own the social world.
Building a social operating system could be a stunningly complex task. Take a photo app, for example. Each photo came with potential privacy restrictions: in order to keep its pledge to users that they controlled who saw their information, Facebook had to maintain restrictions at every step: was this photo available for everyone to view, or just friends?
But now Facebook was promising outsiders that they could make their own photo apps, or anything else, getting the same information that Facebook itself had for its in-house apps. That was part of the appeal to developers. But could those outsiders be trusted with the information?
Max Levchin, a former PayPal executive who had started a company called Slide, felt that such information sharing would be the essence of the Facebook operating system. He had been lobbying D’Angelo to give developers maximum integration with Facebook. That presented a privacy issue. The very definition of a social app requires that the developer get not only a user’s personal information but details about the people they’re socializing with. Because users would, in effect, be exporting their social network, by necessity some of that data belonged to other people. These “friends” of the user who actually signed up for the app might not know that their information was transferred. Should they have a chance to vet this exchange? Furthermore, users might have labeled some personal data as restricted. How would Facebook ensure those restrictions would be honored when developers got access to all that information?
Zuckerberg understood that Facebook had to honor the trust of its users. But he also believed that the social apps that would emerge would be worth the risk of leaked information. “There was a lot of deliberation about which data to share,” says an executive at the time. “There was a very strong thing coming from Mark that was, We need to be able to make it so that other developers can build things as good as what Facebook can build. Facebook was a small company at the time, so they needed to give developers this data to just make the platform desirable.”
Facebook did take steps to prevent leakage. Generally, it required developers to keep certain information in temporary memory caches rather than downloading to permanent storage. And developers would pledge to Facebook that they would not sell or release that data to others. That would be the absolute worst case.
Ultimately, the safeguards were built on an optimistic view of what developers might do. Facebook’s executives at the time now admit that the protections were relatively weak in part because the data held by Facebook in 2007 wasn’t seen to be as critical as it would later be. The stakes were lower and the norms were different. In that time period, the tech community was urging Facebook not to lock down its information but to be more open. Facebook, said its critics, was a “walled garden.” This was the term used when the owner of an online destination owned all the services and features that people used when they visited. These digital “company towns” ran counter to the democratic ethos of the Internet. They smothered innovation. Tearing down the walls of your garden meant you were being a supporter of the free Internet.
So the next great project after Open Reg and News Feed would be Platform. It would cement Facebook’s status as the dominant company in the social-networking world. It would give Facebook a huge edge over competitors. (And help it surpass MySpace, which was already hosting third-party apps.) It would make millions of dollars for creators of popular apps. By allowing others to make use of the accounts of millions of its users, Facebook would become the de facto global arbiter of people’s online identities. And the influx of new users, and the increase of time spent on Facebook, would bring in the revenues it had yet to realize.
Much of that happened, and less. Platform’s legacy also would result in frustrated developers, angry users, and, ultimately, the worst catastrophe in the company’s history.
THERE WAS MORE than the usual move-fast imperative for Facebook to introduce its platform in a hurry. That January, Apple CEO Steve Jobs, to astonishment and acclaim, had introduced the iPhone. The announcement had created a frenzy, and people marked their calendars for the time in June when they would actually be able to buy one.
In theory, the iPhone would not provide competition for Facebook’s platform. Steve Jobs had brushed off criticism that Apple was not allowing software developers to write applications directly to its operating system. In any case, Apple wanted nothing to do with social networks.
But Facebook was wary of Jobs’s intent to close the iPhone to software developers. As a student of Jobs, Dave Morin had seen Apple go to market with a strictly focused product that later took on new powers, hitting competitors with a delayed punch. The iPod was out for two years before the iTunes store.
So Facebook set an ambitious May 24 date to unveil and ship. And it rented the San Francisco Design Center, a large venue in the South of Market neighborhood where start-ups abounded, so that it could invite almost a thousand people for what would be its first developers conference. It called the event F8, a reference to its frequent all-night hackathons where engineers would spend eight or more hours blasting away on a blue-sky idea. Maybe it was a coincidence that it also invoked the word “fate,” implying an inevitability to Facebook’s impending dominance. Maybe not.
In the weeks before the announcement Facebook gave a select group of developers an advance look at the platform so they could prepare their apps to be ready at launch. Some of them had been building widgets—small applications—on MySpace. Others were well-known software companies. Morin learned that Amazon was working on a digital reading device, to be called the Kindle. He tried to convince Amazon to work with Facebook to make it run as a social application, but had no luck. But as a consolation prize Amazon agreed to release an app called Book Reviews, where Facebook users would share their reading experiences. Amazon had no desire to actually write the app, so Fetterman and Sittig whipped it up.
Microsoft and The Washington Post were also launch partners. But Facebook’s favored app in the first batch was a collaboration between two old friends: Joe Green, who had been friends with Zuckerberg since Kirkland House, and onetime president Sean Parker. The two were building a website to empower activists using social networks. When Morin asked them if they wanted to do a version for the platform, Parker saw an opportunity to weave it so deeply into Facebook that people would think it was part of Facebook. “It should feel like a feature of Facebook,” he said. Its code name was Project Agape, but wound up with the name Causes, because Parker wanted it to evoke other official activities on Facebook like Groups or Events. The Causes team scrapped their website, choosing to run directly on Facebook. Zuckerberg liked it so much he offered to buy it for 1 percent of Facebook. “I was like, Okay! But Sean didn’t want to sell—he already had plenty of Facebook stock,” says Green.
All told, seventy developers would be ready with apps when Platform launched. And they would be part of a spectacle that helped change the world’s view of Facebook.
Normally Facebook released product in the dead of night, with at most a blog post to mark the occasion. But Platform was to be Facebook’s symbolic ascension to the top of the tech food chain, a signal that Mark Zuckerberg’s dorm-room slouchers were graduating from the Crimson to the big-boy business pages.
In brainstorming the event, Morin had a single template in mind: Steve Jobs’s celebrated Apple keynotes. To produce the graphics for Mark’s speech, Facebook used Ryan Spratt, who had worked so much on Jobs’s slides that Apple eventually gave him an office. To help conceptualize the message, Morin tapped Stone Yamashita Partners, a consultancy with extensive Apple experience.
All of this would require something that Mark Zuckerberg had never done: keynoting a glitzy public event. Zuckerberg, of course, couldn’t be expected to match the glib elegance of Steve Jobs. “He’s an amazing communicator now, but in that time period he was still learning,” says Morin, perhaps overly kind in his current assessment. For Zuckerberg, the stress of speaking triggered unusual amounts of perspiration. In coming years, he would demand that the backstage area at his speaking events be cooled to well under 60 degrees. Brandee Barker would often wind up blow-drying his armpits before he went on stage.
In the course of brainstorming concepts for his speech, Zuckerberg acquired some of the language that would pepper his explanations of Facebook’s mission for years to come. The most important term was what he called the “social graph.” Though the concept had been batted around for months in late-night discussions—and was pursued as far back as Adam D’Angelo’s Buddy Zoo—that single term seemed to embody what Facebook wanted to unlock for its users.
Social graph refers to the nexus of connections people have in the real world. By expediting connections to those people who were on your friend-and-acquaintance radar, Facebook was unlocking a network you already had, keeping you in close touch with people huddled next to you on this virtual constellation, and drawing lines to those who were one, two, three degrees away.
“We don’t own the social graph,” Zuckerberg would explain to me later that year, going slow so even a mainstream journalist might understand this dive into network theory. “The social graph is this thing that exists in the world, and it always has and it always will. A lot of people think that maybe Facebook’s a community site, and we think we’re not a community site at all. We’re not defining any communities. All we’re doing is taking this real-world social graph that exists with real people and their real connections, and we’re trying to get as accurate of a picture as possible of how those connections are modeled out.”
Once that picture was captured, Facebook and all the other companies on the platform could exploit the social graph to, as Zuckerberg puts it, “build a set of communication utilities that help people share information with all of the people that they’re connected to.”
Unsaid was Facebook’s ambition to be the only company that captured the full picture of the social graph. It would be like a search company having exclusive access to the World Wide Web.
For weeks, Zuckerberg rehearsed his speech over and over, down to the hand movements and where he would talk on stage. But he’d still be his authentic self, wearing his signature fleece and jeans. That extended to his feet, which would be shod in the unfashionable flip-flops he wore just about everywhere. At the last minute, he discovered the Adidas sandals he favored had been discontinued and an assistant had to scramble to find a pair still on sale. (They bought ten, to stock him for the future.)
As much as he hated public speaking, he knew he had to do it. Zuckerberg would enter the arena, a Cicero of the software world, and explain how Facebook would create the next great platform. His first line would set the tone, and past midnight on the eve of the speech he was still rehearsing, intoning the line over and over.
Today, together, we’re starting a movement.
At 3 p.m. on May 24, 2007, Zuckerberg took to the stage. All that practice paid off, as he got through the speech with no long silences or flop sweat. In any case, it was the substance of the speech that wound up impressing people. Though Facebook had been welcoming everyone for months, the tech elite had still regarded it as a college site. F8 forever changed that perception. Zuckerberg recited a list of statistics to prove this. Facebook’s user base of 20 million was growing by 100,000 users a day, he reported, with the twenty-five-and-over demographic the fastest growing. It was the sixth most trafficked site in the world. And Facebook was already the most popular photo site on Earth.
Following the event, Facebook hosted a giant “hackathon” where developers could spend the night coding new apps for the platform. Facebook didn’t invent those all-night coding parties, but they dovetailed perfectly with the move-fast company ethos. While coders cobbled apps to run on Facebook, Zuckerberg, D’Angelo, and Moskovitz were in the lobby of the nearby W Hotel, making sure that the system would not crash.
The night before the event, Zuckerberg and Morin had sat on the lip of the stage speculating how many developers the new platform might draw. It was hard to judge. After thirty years, Apple had only 25,000 developers. Google had about 5,000 developers making widgets for its user-customized iGoogle home page. “I remember thinking, If we could grow things at that speed that would be pretty amazing,” Morin says. So his goal was 5,000 developers. He dreamed of it happening in a year.
It took two days.
ILIKE WAS FOUNDED by twin brothers whose parents had come to America as children when their family fled the Iranian revolution. Both Hadi and Ali Partovi had earned computer-science degrees and worked for Microsoft. Then they started a company where people could share their musical preferences with friends and buy things like concert tickets. iLike had existed as a website for a year, and had an app running on MySpace, but was not making much of a mark on the world. When they learned about Facebook Platform, Hadi, the president, urged his brother (and CEO), Ali, to go all in. “In the history of computing, there was the personal computer, there was Windows, there was the Web, and now the Facebook platform,” Hadi told his brother.
They had hardly returned from the launch when the gamble seemed to pay off. In the first day, they got 40,000 users, which roughly doubled the number of people who used the site on a given day. The user base rose into the millions. Each new user was not only downloading the app but also uploading huge databases of music they owned. “That has a tremendous impact on your infrastructure,” says Nat Brown, who was iLike’s CTO at the time.
Ali Partovi called up Morin in desperation and asked if he knew anyone in the Bay Area with extra servers. Morin knew of a company in Oakland, so the iLike guys flew in from Seattle, rented a U-Haul at the airport, and trucked the servers from Oakland to one of the data centers Facebook was using. Morin also arranged cages of new servers for other apps, many of which were taking off like a bottle rocket. Others were also topping a million users. Since Facebook only had around 20 million users at the time, the numbers seemed otherworldly.
So why did the Platform instantly surpass Facebook’s most optimistic guesses? The secret turned out to be that the News Feed was a more powerful engine of distribution than even Facebook suspected. Less than a year after its introduction, Facebook was still tinkering with the algorithms that determined the ranking of possible stories on people’s feeds. The developers were way ahead of them. In order to popularize their products quickly, they had been experimenting in techniques, sometimes dicey ones, to take advantage of the peccadillos of various platforms. They also understood human nature well enough to know why people click on some things and not others. Some had already mastered the mysterious art of “going viral” on MySpace and other networks and knew just what to do to exploit it for their own gains—and the detriment of Facebook’s users.
Two companies with virtual black belts in virality were Slide and RockYou. They had both built huge followings on MySpace. But despite all that engagement—or maybe because those apps had used scorched-earth tactics to grow—MySpace had become unhappy with its developers. It felt that some weren’t adding much value, and others it came to view as competition. “MySpace at that time was unfriendly to third-party developers,” says Lance Tokuda, CEO of RockYou. “In one meeting, Chris DeWolfe actually said that they might kick everyone off the platform.” So when Dave Morin promised them that Facebook planned to give developers that same access to its system that its own engineers had, RockYou and Slide jumped right in.
Slide and RockYou specialized in activities that wasted time. It was almost as if they had a competition to produce the most mindlessly addictive activities. Their first Platform offerings didn’t even bother to build something original, but embellished features that Facebook already offered. Slide’s most popular app was called SuperPoke!, which extended the powers of Facebook’s dumbest feature. Its CEO, Max Levchin, bought the small company that developed the app and unleashed it in the Facebook ecosystem, like introducing invasive Asian carp into American waterways. The theory was that Facebook users were bored with just poking one another and were hungry for sillier means of nudging a friend. The super poke that caught on was “throwing sheep,” which became a symbol of the mindlessness of Facebook apps. (Levchin still defends SuperPoke!, claiming that it added “vibrancy and zest” to Facebook communications. One could also argue that those hurled sheep and such were precursors to the emojis that would rule a decade later.)
RockYou had its own version of Poke called Hug Me. “Hugging was our most popular action,” says Tokuda. “Also, you could smile at, you could dance with—any kind of verb that was fun for users.” The two companies were always trading accusations of ripping each other off.
RockYou’s signature app, though, was Super Wall, which allowed people to replace the Facebook wall on their profile with a gaudier version that let people upload videos and other media. Because Super Wall would work only if your friends were using it as well, RockYou’s distribution strategy was to do all it could to clutter people’s Walls and News Feeds with invitations. “We would get the list of IDs of their friends,” says Tokuda. “And with that list, we could let them invite other friends to invite all their friends to join Super Wall and share content, because for Super Wall we needed everyone connected.”
“It did turn into a Wild West,” says Levchin of the scramble to corral new users. “Companies would compete with each other on who was loudest and who could incent[ivize] the users to share the most.” He admits that Slide was an offender, consciously creating viral loops that would suck in users.
Another company that courted massive engagement was an entertainment app called Flixster. Ostensibly, it was a diversion for movie buffs, creating quizzes they could take to show their cinema knowledge. But that was a subterfuge to draw traffic. “Practically speaking, it was a viral engine to get kids to basically create quizzes and spam their friends,” says Brad Selby, senior product manager of Flixster. “And it worked very well.”
And then came the games, which were in a class by themselves in terms of disrupting the News Feed.
Mark Pincus smelled the social-game opportunity first. He was the guy who, with Reid Hoffman, had been part of the angel investment that, in his words, was like winning lotto. In late 2006, Matt Cohler tipped him off that Facebook was going to launch a platform and was looking for entrepreneurs to come up with apps. We don’t want any money from you, he told Pincus. Just build cool stuff and we’ll expose you to our traffic.
Pincus had already identified that games were the missing piece on Facebook. It was something he had always wanted to do at tribe.net, his own failed social network. He started it now, calling his new company Zynga. “Games,” he says, “were the perfect thing to drop in the middle of this cocktail party.”
Specifically a poker game. What would be more social? “A poker game is like an always-on bar, like Vegas or something,” he says. “You can go online there with your friends, you can meet other people.” Pincus had tried to build online poker on the web, but the technology didn’t work well. Building on Facebook, though, would solve a number of problems: you would know whom you were playing against (because Facebook used real identities), and you could play with your friends.
Pincus met with Zuckerberg often, a lunch or dinner every other month or so. Just the two of them. They became friends, and Pincus would attend things like the parties Zuckerberg had on his birthday. “I was the non-Harvard, non-Facebook person there,” he says. Pincus was in awe of how relentlessly the younger man could soak up knowledge. Zuckerberg was a learning machine. As a poker player, Pincus respected Zuckerberg’s skill at not showing his virtual cards. He would always walk away from the table a winner. But at times when his own interest wasn’t threatened, he would be generous with advice and aid. “He was always being pitched by a lot of people all the time and he had good judgment to pull things out of what you’re telling him that were useful,” he says. “When he said, Okay, I like that idea, you knew he meant it, and you knew he was probably going to do something with it.”
Pincus knew that Zuckerberg personally didn’t think that games were the ideal use of the platform. “Their vision was Causes,” says Pincus. “They felt that [the Platform] was going to bring out our best selves.” But though Causes at first gained a lot of users supporting worthy movements, Causes didn’t make money like silly programs or games that displayed ads on their little slices of Facebook. The investors in Causes, including Bill Gates, would eventually lose their money.
Zynga, though, was a booming business. Hold ’Em Poker was an instant hit, the first of many. People got invitations to join games, and notifications when their friends joined a table. When a wildly popular online Scrabble game was quickly closed down when the copyright holder, Hasbro, threatened legal action, Pincus jumped in with his own version, Words with Friends.
Then Zynga came up with a social game called Farmville. Users acquired livestock, crops, and equipment to tend to virtual farms, infesting the News Feed with invitations and status reports with every new chicken and tractor. It was the epitome of a time waster. It was also a giant money maker. Besides advertising, Farmville reaped revenues by selling virtual goods. People became obsessed with developing their farms and sped the process by purchasing make-believe equipment, seed corn, and even trees. Farmville also pushed the envelope in getting its users to become its pushers. The first thing you did on Farmville was give “gifts” to your friends, luring them into the quicksand of virtual agriculture. They’d learn about the gifts, of course, through News Feed. At Farmville’s peak, 80 million people would become virtual farmers. Eighty million.
With hundreds—thousands—of developers tapping into Facebook’s API to distribute content about their apps, the News Feed became jammed with junk posts, a tsunami that engulfed its normal operation. Users also got bombarded with notifications, which developers also could use to distribute “news” of their apps.
So at the same time Facebook was celebrating the takeoff of Platform it was worried that bad behavior might be poisoning the system. “We’ve got venture capitalists, we’ve got entrepreneurs, and there’s developer events and all these things going on. But this is impacting the users’ experience and it’s spammy,” says Dave Morin. “I believe the word ‘spammy’ literally became a trending word in the world that year.”
At any given time Facebook might show the average user around 1,500 possible stories involving the activities of their friends. Its ranking algorithm would try to narrow that down to a hundred or so. During the average session, users might look at only the very top-ranked half dozen or so. Instead of learning about what their friends were up to, or seeing pictures of cool parties, or learning who was in or out of a relationship, people were scrolling through dozens of posts that someone tossed a sheep, scored high in a dumb quiz, or invited them to play some silly game.
“If you can get a friend to bother ten friends to get one more user on your app, you’re very happy because you got one more user,” says Josh Elman, who joined the Platform team in 2008. “Facebook, though, has nine other people who’ve just gotten bothered.”
Definitely not the revolution Facebook had in mind.
FACEBOOK BEGAN TO correct course, limiting developers’ access to the News Feed and notifications. “The number of developers and the amount of stuff like spam that we had to deal with just grew much faster than we were ready for,” says Adam D’Angelo. “So there was this whole crackdown we had to do.”
Naturally, the developers hated those new rules. Slide’s Max Levchin considered it a bait and switch: Facebook had encouraged them to court user engagement in their tactics. “They said, Go for it,” he recalls now, noting that Facebook itself used engagement as an internal metric, and all that activity Slide created was bolstering Facebook’s business. “One person’s spam is another person’s entertainment,” he says.
But the ones who hated the restrictions most weren’t the ones who wantonly flooded the News Feed. It was the law-abiding developers. They felt that they were being punished for someone else’s bad behavior.
Joe Green of Causes griped to Morin, “You need to punish the wrongdoers.” But that would require Facebook to have its managers make actual judgments on developer behavior. Which was not the Facebook way. The company was operating at a scale where only algorithms or armies could make choices at such scale, and it didn’t want to hire armies. “Facebook didn’t want to have to do things with human oversight—they want it all to be [automated],” says Green. (It would take a long time for Facebook to understand the limits of algorithms and the necessities of armies.)
The rule changes dampened excitement for many developers who had bought into Facebook’s promises that Platform was the next Silicon Valley gold rush. Thousands of entrepreneurs had been starting up companies on the idea that Facebook’s Platform would be like the web, supercharged by the juice of social activity—a must-do for any business. Now it wasn’t so clear.
iLike was one of the hardest hit by the crackdown. It had been the most popular Facebook app; its CEO told The New York Times of dreams that it would be “the next MTV.” Though it did use the News Feed for things like telling the user’s friends that he or she scored high on a quiz, says Nat Brown, iLike’s chief technical officer at the time, it didn’t make stuff up to get on the News Feed like other apps did. “We felt like we were at a disadvantage because we were more respectful of the rules than other players,” he says. “We were this great place where users are really interested in music, but they’re saying all apps are bad because RockYou messages their friends 100 times an hour.” With its access to notifications and the News Feed curtailed—Facebook used the term “deprecated”—iLike’s growth hit a wall, and it began a slow decline. And then a rapid one.
“There was no way to maintain the business we had built on our Facebook app, and so it became clear to us that whatever business we had in our Facebook app was short-lived,” Ali Partovi would later say in a deposition. In 2009, iLike, which had once gathered tens of millions of Facebook users, sold itself to MySpace for the fire-sale price of $20 million.
“Facebook is a rocket ship,” says Nat Brown. “It turns out iLike was not strapped to the rocket ship. We were the fuel.”
THE NEWS FEED spam wars were only the start of a push-pull between Facebook and its developers. Facebook would change the rules and developers would figure out how to get around those rules. The developers would share techniques with one another. When they tried a particularly sketchy feature, they learned not to show it to anyone they identified as a Facebook employee, or would use geo-tagging to exclude the Bay Area. “We were playing a little bit of a cat-and-mouse game and a lot of times I think we were behind the mice,” says Facebook’s Josh Elman.
One of the more serious instances of misbehavior came when developers sold space on their pages to low-quality ad networks. Premium advertisers usually weren’t interested in the mind-numbing apps that ran on Facebook. The companies buying ads were often bottom-feeders, involved in the dicey practice of lead generation, where they would use deceptive tricks to try to get people’s money or data. An example would be ads enticing a user to click, instantly installing a browser that would thereafter stealthily scoop up all of the user’s subsequent web behavior. Getting rid of it would almost require a computer-science degree.
In a 2009 exposé of these practices, TechCrunch described how the lead-gen advertisers would abuse Facebook customers who fell for offers of low-value items like game currency, introductory offers for services, and other trifles. Facebook’s Platform, it joked, should be dubbed “Scamville.” Its author, TechCrunch co-founder Michael Arrington, described one scam where users are asked to send their mobile phone numbers to get the results of a quiz. When the text arrives, it gives them a pin code to enter for the score. They didn’t know it, but they just signed up for a service that charges them ten dollars a month.
Arrington noted that while Facebook had policies against these abuses, “those rules are routinely ignored by developers, and are rarely enforced by Facebook.” (He also cited MySpace as the host of similar abuses.)
Zynga’s products were among those hosting deceptive ads, but Pincus says it wasn’t his fault, noting that the ads were automated. “We had no control over what the [advertisers] put in,” he says. “We would get paid if our users did something.” Besides, he said, the same ads would show up on Google. “We were being held to a higher standard.”
But Pincus didn’t help matters when he spoke to a group of tech founders at a small Berkeley gathering. “I knew that I wanted to control my destiny, so I needed revenues, right fucking now,” he told the young engineers. “So I did every horrible thing in the book just to get revenues right away. I mean, we gave our users poker chips if they downloaded this zwinky toolbar . . . I downloaded it once and couldn’t get rid of it. We did anything possible just to get revenues so that we could grow and be a real business.”
Pincus now says that he was exaggerating, mouthing off to a bunch of aspirational founders over drinks. So why didn’t he refute his statements publicly? Because, he says, he didn’t want anyone to know how he was really making money. “My customers were middle-aged women in Indiana who’d stopped watching soap operas to play Farmville. Some of them were spending huge amounts, thousands of dollars a month with us. But I didn’t want to get that story out. So I had to take the arrows, because I was okay with the rest of the world thinking we’re making [money from scamming].”
But though Pincus had lit on a perpetual moneymaking machine, he was caught in a strange dance between his company and Facebook. For Zynga, access to Facebook became an existence question when the company cut back on spammy distribution. Facebook’s solution: buy ads from us. Pincus ponied up, becoming Facebook’s biggest advertiser. Without steady access to the News Feed, its main pipeline became the ads on the “left rail,” the screen space alongside the feed. Two-thirds of its traffic came from people clicking on ads.
By then Facebook was pressuring Zynga by other means: in 2010, the company had introduced its own form of in-house currency called Facebook Credits. It urged that developers use this form of payment, which kicked back 30 percent of every transaction to Facebook. “We had real issues with Credits,” says Pincus. “Number one, credits sucked. We were testing Credits for them versus PayPal and we had huge losses on anyone we put through Credits.” The second reason was the unfairness that Facebook was forcing Zynga to use Credits, while the choice was voluntary for other developers.
Pincus came in to talk to Zuckerberg, who brought in Sheryl Sandberg, saying that since she had worked for the Treasury Department and understood economics so well, she could clear things up for him. “They had this whole argument—they felt like we’re the biggest user and we were somehow being subsidized where we were using more than our fair share,” says Pincus. “She had this whole explanation that it’s the tragedy of the commons and all.” But Zynga decided not to comply if others didn’t have to. “I said, Fuck that,” recalls Pincus. “When you make it mandatory, I’m in. But until you do, I’m not.”
Pincus considered himself a friend to Zuckerberg. He respected Sandberg. But he knew that ultimately it was each side for themselves. “They are amazing people because they’re tough as nails but also lovely and nice,” he says. “Like soft boxing gloves with brass knuckles. You do not want to get in a scrap with either of them. And I did.”
Zynga began exploring alternative venues to Facebook for distributing its games. For a while it was a vicious standoff. In trying to negotiate terms by which Zynga might stay, Facebook offered a contract that Pincus would not sign. Among the provisions were that Zynga could not move its games to other platforms. “We held out and Facebook was getting more and more angry about the user feedback from people who didn’t want to see games in the feeds, he says. “We weren’t abusing anything, we were doing what Zuck had told us.” Pincus even began talking to Google as an alternate partner. Meanwhile, his team was frantically coding up a separate website to host the games if Facebook tossed them off the platform.
It was only Pincus’s friendship with Zuckerberg that led both companies to back off. The two had a series of meetings, some going as late as 4 a.m. “He was a night owl who could drink Diet Cokes all night,” says Pincus. “He said, Look, there’s no one in a position to compete with Facebook. You and Google are it.” So each was a threat to the other. They reached a complicated deal where each side got something, but, as Pincus explains, things played out more loosely in practice. “We averted nuclear war,” says Pincus. They signed the deal in May 2010. For a few more years things thrived.
“At one point we were eighty percent of their API usage,” says Pincus. “At our peak with them, we were sixty percent of their app DAUs [daily average users]. And I heard that by the time they went public, we were something like twenty percent of Facebook’s overall revenues.” Facebook was so dependent on Zynga at that point (its 2012 IPO) that the prospectus listed it as a business risk.
Nonetheless, tensions still remained, and as smartphones became ubiquitous, Facebook’s platform became less valuable to Zynga. “It was clearly all going to be mobile and it didn’t matter anymore,” Pincus says. In 2012—three years before the five-year timeframe of their contract ended—the two companies renegotiated. Zynga would no longer be a Facebook-first partner. It was a symbolic moment of the dashed dreams of Platform.
“I naïvely thought Facebook would see that their users got the most value,” says Pincus now. “They’d stay with Poker, Farmville, and all these things the longest, and Facebook would make more economic value and so they would want to promote our games, and they didn’t. They were an ad business.”
Pincus, of all people, should have known better.
In a sense, Pincus and the other developers who wrote applications inside Facebook were already fighting the last era’s war. Only a year after announcing the original Platform, Facebook in effect created a new way for developers to access information from Facebook—and for Facebook to bring software companies into its ecosystem. Called Facebook Connect, it allowed developers to use Facebook as a log-in on their own services and apps. The applications would live outside of Facebook. It was sort of a resurrection of Fetterman’s original API idea, living side by side with the Facebook platform.
Mike Vernal, yet another engineer who had come over from Microsoft, headed the project, which had two purposes. First, it was meant to solve the problem users had of trying to create and remember a login for every online service or site they signed up for. “I feel like I should have one and be able to log in everywhere too,” says Vernal. Also, he adds, “We thought that a bunch of apps and industries could be fundamentally better if they were more social.”
Facebook Connect was a step toward making Mark Zuckerberg’s company the de facto arbiter of identity on the Internet. Your Facebook persona could be used on thousands of other sites. And since you were logging in with Facebook, Zuckerberg’s company would be able to monitor your activity.
Facebook already had thousands of developers, but this would raise the number dramatically. And Facebook would also be sharing information it had about users (who intentionally signed up for the apps using Facebook Connect) and friends of users (who had no idea that their information was being passed on to apps that they might never have heard of, let alone signed up for).
What data Facebook gave the developers was supposedly dictated by its rules. But according to some developers and emails that later were exposed due to legal actions, it turned out that in practice those rules were flexible, that a bartering situation emerged when it came to what personal user data was supplied to developers. “There were nominal guidelines, but that was complete nonsense,” says Selby of Flixster. “It was absolutely catch as catch can—who can you convince. One week, we’d say, You know what, we could really do something with a lot of friends-of-friends data about movie likes. And they would go back in their darkened room and say no. And we’d say, Let’s pitch it to you differently—if you give us this data, we project we’ll increase engagement, and it’ll flow to you. And they might say, Okay, that makes sense, and they’d flip the switch. Or they might say, Jump in a lake.”
In the short term, at least, Facebook was motivated to let the carnival continue. Because if all the developers writing Facebook apps left the Platform, there would be a lot less traffic on Facebook. “What was in it for us was very simple,” says Dave Morin. “It was creating more time, and more inventory [for ads]. One of the things about Facebook that’s always been very straightforward is, we create experiences that are highly engaging, the business model is ads, and so the more engaging, the more ads, right?”
Some executives at Facebook warned that allowing all the junk posts from developers would begin to alienate actual users. Will Cathcart, an engineer who left Google to join Facebook in 2010, dove into the data and found an alarming trend. “One of my growing fears is that we’re routinely erring on the side of avoiding pain for developers and in the process causing user pain,” he wrote in a 2011 email. He cited data to indicate how people were tiring of developers’ tricks. “Users don’t trust apps to do the right thing,” he wrote. Furthermore, users didn’t trust Facebook to do anything about it. When they reported bad behavior to Facebook, they felt that nothing was done. Cathcart said that people he knew personally had concluded that flagging violations to Facebook was useless, so they stopped doing it. The reply from his boss, Mike Vernal, was that . . . it wasn’t so simple. “This is tricky stuff,” he wrote. “One week everyone is yelling we’re not protecting users enough. The next week everyone is swooping in and saying we’re being too aggressive. It’s a delicate balance but both sides are right.”
His suggestion: be careful about sanctioning developers. “We need to soften the punishment ASAP so we can protect users without screwing developers.”
BY 2010, IT was clear that the Platform needed a fundamental rethink. Hundreds of thousands of developers were using Facebook Connect, but those writing applications on the canvas pages of Facebook were lagging. Facebook made one more adjustment, creating a new API that allowed for even deeper integration into its system.
Zuckerberg had always seen Platform as a way to extend his worldview of sharing to a larger audience, and felt that this would help. Seven years into the Facebook experiment, he now believed more than ever that people would be better off if they knew what their friends, families, and contacts were up to. “Mark has a distinct time in his life when he started talking about sharing information and understanding what your friends were doing,” says Don Graham, who besides being CEO of the Post was on Facebook’s board at the time.
Zuckerberg’s new buzzword was “Open Graph.” Just as the social graph mapped your personal network, the Open Graph would map the interests and activities of those you knew. Maybe, by noticing your affinities with casual connections on the Open Graph, you’d get closer. Or maybe you’d just learn more about people you knew.
Zuckerberg announced the system in 2010—Graph API V1, the first version. A year later, he was still talking excitedly about it. He explained it all to me one summer day in 2011, not long before the September F8 conference. We were walking in College Terrace, a leafy Palo Alto neighborhood where Facebook had its headquarters at the time.
That year Facebook would be introducing some new tweaks that really showed how user information from apps could be shared with Facebook. The key partners were the music-sharing system Spotify, the video-streaming platform Netflix, and The Washington Post, which developed an app called Social Reader. The products introduced were not really stand-alones, but social extensions of their main applications—a means to allow users to circulate to their personal networks what they were listening to, watching, and reading. The idea was that eventually, every application and service would have a companion app on Facebook that enabled people—presumably with approval—to share their exercise routines, media preferences, and purchases. Zuckerberg would soon predict that in five years the top 100 mobile apps would be part of the Open Graph.
That struck me as a potential nightmare of personal transparency. I tried to come up with an example for Zuckerberg. What if one of his employees called in sick—while Facebook was reporting that they were binge-watching Breaking Bad?
“I’d ask how they were feeling,” he told me.
Originally, it seemed like those key partners were doing wonderfully. But it turned out that it was something of a replay of the algorithmic overkill of the original Platform, where people’s News Feeds piled up with reports of what people were doing on those partner apps. “We couldn’t believe how many people adopted it, and how much they liked it,” says Graham of the Post Social Reader. “That was the problem. Everybody’s page began filling up with everything everybody read on social media—the Facebook algorithm overweighted them. Then Mark and Chris [Cox] didn’t like the outcome and started under-weighting it. At that point, it didn’t collapse but it sucked.”
None of the original applications in that generation performed the way Facebook envisioned, and the tsunami of apps it hoped would follow—apps where people shared fitness data, location data, and other information—did not become part of people’s lives the way Facebook had hoped.
By then, it hardly mattered, because developers had found a much better operating system—two of them, in fact. Apple and Android had created their own development platforms for their mobile phones. Developers quickly understood that mobile was the best place to build their businesses.
Facebook’s original ambitions for Platform—a thriving operating system where developers would write original apps that ran inside Facebook—were over. “Unfortunately, mobile just completely undermined the entire system and basically relegated the platform to irrelevance,” says Facebook’s head of partnerships, Dan Rose.
Facebook’s scaffolding of its platform still remained, and for various reasons, developers still built apps that could be deemed as social, or at least made use of the socially oriented information shared by users. And Facebook Connect—which worked fine with mobile apps running on Apple or Android—remained wildly popular. The reason was simple: being a Facebook developer gave you the access to Facebook data that could add social juice to whatever you were doing in the first place.
Sam Lessin, Zuckerberg’s Harvard classmate who joined the company in 2010, put it to Zuckerberg this way, in a 2012 email exchange about the future of Platform:
Right now I believe if you asked an application to implement Facebook Connect but didn’t give the friend graph . . . people would have no reason for implementing it at all.
Facebook wanted to make sure that it got information back from that exchange as well, and in 2012 it drew up a tougher deal. Introducing Platform 3.0, it decided to ask for what it called “full reciprocity” from developers. In exchange for Facebook’s data, developers had to share with Facebook the user data that they gathered. Mike Vernal put it this way in an internal chat: “When we started Facebook Platform, we were small and wanted to make sure we were an essential part of the Internet. We have done that—we are the biggest service on earth. . . . Now that we are big . . . we need to be thoughtful about what integrations we allow, and we need to make sure we have sustainable long-term exchanges.” Translation: we may not ask developers for money to get our information (though Zuckerberg considered that, and Facebook executives discussed it extensively in that period), but we need something. Like your data.
In an internal email explaining it, Zuckerberg wrote:
We’re trying to enable people to share everything they want and to do it on Facebook. Sometimes the best way to enable people to share something is to have a developer build a special purpose app or network for that type of content and make that app social by having Facebook plug into it. However, that may be good for the world but it’s not good for us unless people share back to Facebook and that content increases the value of our network. So ultimately, I think the purpose of the platform . . . is to increase sharing back into Facebook.
Zuckerberg made it clear: by now Platform’s key attribute was a means of information exchange between Facebook and developers—the exchange of user data, where the user had little awareness of how his or her personal information was being shared. Reciprocity aside, by far, the biggest movement of data was from Facebook to the developers.
Facebook would not let that stand. As a cache of documents later revealed, Facebook in that time period was blatantly planning to limit or ban developers that it considered potential competitors, and deny information to those that weren’t delivering value back to Facebook. Zuckerberg yanked API access to a contact management start-up called Xobni; when Facebook began thinking of creating its own Gifts feature, it pulled support for an Amazon Gifts app it had already approved. And in 2013, Facebook began to consider a more sweeping adjustment that would more generally curtail the widespread giveaway of friend information, smashing the business plans of many companies that had built social apps.
Zuckerberg’s original vision had been to give outside developers the same access to tools and the News Feed as Facebook had to develop its own features. Now those software companies who had invested in that dream were being shut out. Facebook had turned its promised “level playing field” on its side.
While most of the executives and product managers followed Zuckerberg’s lead, there was some grumbling, in particular by Ilya Sukhar, who had come to Facebook when it bought his company, Parse, which made tools for developers. Standing up for those stakeholders, he was feeling lonely. “I feel like I am the only one with a principled stand here,” he wrote in an internal chat thread among officials in October 2013. “I just spent the day talking to many dozens of devs that will be totally fucked by this and it won’t even be for the right reason.”
The right reason, of course, would have been to close the Friends API because it gave developers the personal information of users who were unaware of the exchange—and Facebook could do little to control that information once it left Facebook’s servers. “We were creating user experiences that were just awful from a privacy experience perspective,” says one executive on the Platform team. “You would log in to these apps with Facebook and suddenly they knew everything about you and your friends. And they were doing very nefarious things with it.”
Now Facebook would stop the practice, not to serve users but because it did not want to give away data to developers for nothing in return. This was not a friendly message to share at a developers conference. So Facebook came up with the idea to announce the change as if it were motivated by concern for user privacy. The move would fit in with a set of privacy features already planned for release. One executive dubbed the PR tactic “the switcheroo.” The PR people helped shape the announcement—to be made at the April 30, 2014, F8 conference—to lead with the idea of “giving users more control.” And so, despite Facebook’s self-serving motivation for what it internally called the “friend deprecation,” Zuckerberg spent the first part of his keynote explaining how Facebook was standing up for privacy by, among other things, shutting down Graph API V1 and introducing V2, thus ending friend-of-friend access.
But Facebook allowed some developers to avoid the lockout—if they kicked back data or committed to buy ads. It put those favored developers on a white list that still allowed access to friend data. The white list included big names like Apple and Netflix. The bartering could get imaginative. To solve a trademark dispute with the dating app Tinder, Facebook apparently gave the service “full friend access.” At one point, Zuckerberg floated the idea of giving full friend access to game developers who kicked back 30 percent of their revenue to Facebook.
Other developers managed to keep the information flow going by committing to what was becoming a significant part of Facebook’s revenues: a program called NECO where developers would pay for “app installation ads” on Facebook. The Royal Bank of Canada, for instance, got access to the extended API by promising to pay for “one of the biggest NECO campaigns ever run in Canada.”
Facebook made one concession for developers victimized by the switcheroo. Developers would have a grace period of one year before it would block off access to the friends API. Between April 2014 and April 2015, Facebook, despite Zuckerberg’s boasts that the new version of Open Graph was closing a privacy hole, would allow the practice to continue.
“In retrospect, I think we should have given a ninety-day or thirty-day notice, and just moved faster,” says Mike Vernal now.
It is ironic that this one sop Facebook gave to its stranded developers would become a critical factor in the biggest scandal in the company’s history. One might even call it karma. But Facebook would not learn that for four more years.