The photos and updates began appearing on individual profiles, then popped up across interrelated news feeds: The first Facebook-staffer wedding had gone off without a hitch. Two beloved engineers, Ruchi Sanghvi and Aditya Agarwal, had arrived at Facebook as a couple in 2005 and survived the unique pressure of cranking out code for the hottest startup in the world. “They were a package deal,” says Facebook founder and CEO Mark Zuckerberg. So, over what the company still calls its “Christmas break,” Zuckerberg and more than a dozen past-and-present Facebook indispensables — including now-departed cofounders Adam D’Angelo and Dustin Moskovitz — trekked to a beach in Goa, India, for a week-long family celebration. Everyone dressed in costumed splendor; Zuckerberg looked fetching in a maroon silk sherwani. Women flashed henna tattoos. The groom arrived on horseback.
The elaborate Indian ceremony, a Bollywood spectacle with a big helping of Silicon Valley, presented a rare, vulnerable moment for the Facebook infrastructure — one rogue wave could have taken out much of the site’s brain trust. But it also offered a point of reflection for the crew. “All of us together, in that beautiful place,” one attendee recalls. “We’ve come so far. Literally.”
Facebook, which is just turning six, has achieved a level of maturity most wags thought would never come. Somewhere along the road to becoming the platform of choice for 400 million users in every country on earth, the company grew up. Baby photos now dot the worktables at its Palo Alto headquarters. Chefs provide free gourmet fare in the company cafeteria. And the founder, who once coded the site while dashing between makeshift offices in a beat-up car that didn’t need a key, now mingles with his 1,200 employees, recruited and supported by a real HR person, in a new 135,000-square-foot office space. “We used to stand outside of Stanford looking for engineers to help us,” laughs Chris Cox, vice president of product, and creator of the original news-feed feature.
Today, Facebook feels the way Google, Intel, and Microsoft likely did at similar stages in their own life cycles — still agile enough to invent the future, but sufficiently stable to handle some real turbulence. In fact, Zuckerberg has been studying those companies, and their histories, closely. “There are advantages to being both bigger and smaller,” he tells me. “But the cool thing is, we’re in our sweet spot now.”
Zuckerberg himself, a majority stakeholder who cannot be shoved aside, exemplifies the way that crushing the competition has freed the company to gamble even harder. “A lot of companies are set up so that people judge each other on failure. But I’m not going to get fired if I have a bad year. Or a bad five years.” It’s an attitude he hopes will outlast him, and will liberate even those who are less insulated. “I don’t worry about making things look good if they’re actually not,” he says, nodding to invisible investors in the room. “So many businesses just get so worried about looking like they might make a mistake that they get afraid to take any risks.”
It is inside Facebook’s engineering ranks that the company’s love of risk really reveals itself. “We move very fast,” says Andrew “Boz” Bosworth, an engineer who heads up Facebook’s “boot camp,” a six-week introduction into its code and customs that he calls a “cultural indoctrination” for incoming engineers. “And we definitely fight. We expect people to be passionate, and they’re going to fight to make their case.” In fact, “fight,” along with “entrepreneurial” and “impact,” are words one hears constantly on the Facebook campus. Bosworth arrived via Harvard — he was Zuckerberg’s teacher in an artificial-intelligence class — and Microsoft, an environment he suggests is less than inspired. (“It’s a world like ‘insert product specification, receive code,’ ” he says.) At Facebook, by contrast, “it’s free form. If you’re not coming up with new ideas, then you’re just along for the ride.”
The sparring — which takes place at meetings, in prototype demos, and in constant online conversations, and often gets mischaracterized as staff revolt — focuses routinely on the small issues, such as how to make the main page less cluttered. But the big-picture stuff also looms large. “I didn’t realize it from the outside, but the change in going from a [college-student-only] site to being open to the world was extremely controversial,” says Paul Buchheit, who came to Facebook last August after his company, FriendFeed, was acquired. “Most people [at Facebook] thought it was a bad idea and was going to ruin the site.” Buchheit, a Valley legend, was an early Google employee who helped develop Gmail and coined the Google motto “Don’t be evil.” He sees Zuckerberg’s decision to push Facebook ahead — despite a big buyout offer from Yahoo sitting on the table — as a defining moment. “They basically gambled the whole company on that one step.”
Facebook recruitment videos like this one give prospective employees a window into the company’s culture.
Now at work on next-generation infrastructure at Facebook, Buchheit is a living cautionary tale about how hard it is for big companies to scale. He cofounded FriendFeed after a short post-Google recovery period. By the time he left Google, “the pipeline for releasing features was getting really long. If I checked in code now, it might be live to the world in three months.” It felt good to move quickly again. “At FriendFeed, if we had an idea, we could make it live right away. That’s how I enjoy developing things,” he says. The benefit to waiting is a stable system. The cost comes in time to market, in the ability to deliver even an early version of a future killer idea. “People underestimate the cost of slowing things down and focus instead on the benefits of the increased stability,” Buchheit says. “But you think differently about a product that takes so long. If you’re not sure if it’s a good idea or not, you are reluctant to try it out.” Facebook has weekly product updates, but if you really want to test something, you can push your code out daily to a group of users to test. “We win our fights through prototyping,” says Bosworth. “We get our ideas out there.”
Zuckerberg, like Buchheit and Bosworth, is keenly aware of the importance of rapid deployment and iteration, even as the company has become too big for staffers to shout updates to one another across the room. At the heart of the process is the notion of “hacking,” which Zuckerberg insists is not about breaking and entering: “It’s about being unafraid to break things in order to make them better.” Buchheit, a strong internal voice for hacking, describes it as a mix of arrogance and curiosity. “The root of the hacker mind-set is ‘There’s a better way,’ ” he says. “Just because people have been doing it the same way since the beginning of time, I’m going to make it better.” After years of making and remaking the site, pride of ownership takes a backseat to the sheer rush of creation. “It’s like castles in the sand,” designer Soleio Cuervo says of the countless product changes that happen over the course of a week, month, and year. “What we make won’t last, but we make things fast and get to test our ideas quickly with real users. We’re in it for the impact.”
Determined to keep that mind-set alive as the company grows, Facebook has raised the all-nighter to an art form. “Hackathons,” which started when the site was just a handful of friends around a dining table, are now all-hands meetings held every other month or so. Any project, any idea is on the table. If you can find some friends to work on it with you, go for it. The company provides food, music, and beer. It sounds like so much code-boy BS, except that most everyone shows up, even the lawyers. Even Zuckerberg. And the sessions have produced an astonishing array of popular site features, including video messaging and chat. Bosworth’s boot-camp training comes in handy for the hackathons, since at any given time some 40% of engineers are brand new. “We get them working on every area of the site, so they know different code, different products,” he says, adding that ultimately, cross-disciplinary teams naturally form around good ideas. “If you can get it into prototype, we can evaluate it.” The boot-camp system also yields a healthy measure of empathy for just how hard others are working. “It’s easy to think you are the only one doing something important if you don’t know how tough everyone else’s projects are,” Bosworth says.
New ideas, better products, and more users are still on Zuckerberg’s mind for 2010. There are more mundane chores as well: The site is so large now that the Facebook crew is in the midst of cleaning up years’ worth of the underlying code — a multigenerational hodgepodge of so-called spaghetti, which could pose a threat to site stability and product development down the road.
Yet the real target for 2010 may well be Google, for which Zuckerberg has always professed a profound admiration. After I ask about the big, hairy goal for user growth I’ve heard he has set for 2010, he refuses to put a number on it. “We don’t like to talk about that sort of stuff,” he says. “Three-hundred-and-fifty million is good. But when you compare it to everyone in the world, it’s just a start.” Nevertheless, he can’t help but mention Google and its 800 million users. Shooting for that? Smile. No comment. But it’s the number that hides in plain sight for everyone at the company. “We always have a big dragon to slay,” one staffer tells me, “and this year it’s Google.”
Even more than an assault on a hero-competitor, Facebook’s growth is a shot across the bow of Dunbar’s number, a wonky theoretical limit to the number of people one can supposedly maintain stable social relationships with — like how a dinner-party conversation becomes unwieldy with too many people at the table. Dunbar’s number is brought up at Facebook nearly as often as “risk” and “impact,” and all three concepts are inextricably linked: Can the site maintain its groove under the weight of all those users? “We really believe that if the world becomes more open and transparent, things will be better,” says Zuckerberg, reverting to tech evangelist. Then he mentions Facebook’s engineer-to-user ratio, a point of serious hacker pride. “We’re at the spot where each engineer is personally responsible for at least a million users,” and growing, he says, tracing an upward line in the air. He gives me a “That’s impressive, right?” look before pointing out that Google needs 10,000 engineers to serve its 800 million customers — a much lower ratio. “Yeah,” he smiles. “I think the next few years are going to be a very fun period for the company.”