Fail Harder, Move Fast & Break Stuff, Don’t be Fearful

Mike SchroepferFail harder, move fast and break things, don’t be fearful.

Three phrases to sum up Facebook’s approach to innovation, according to the company’s VP of Engineering Mike Schroepfer (@schrep).

In its short life — just six years old — Facebook boasts some extraordinary statistics. With around 500 million users, its population is greater than that of the U.S. In fact, if Facebook were a country, it would the world’s third most populous after China and India. And its users hang around, Schroepfer said. They spend more than 700 billion minutes a month logged on. Thirty billion pieces of content are shared a month, and two million sites use its plugins.

But for many large companies, Schroepfer said, “success becomes a break on innovation.” Speaking at the Science Gallery (@sciencegallery) this evening, Schroepfer said Facebook still innovates “despite our own success.” His comments at the sold-out event were part of the Innovation Dublin festival, and came on the same day Facebook announced its new email service.

Fail Harder: Experimentation is key to true innovation, he said. In 2008, Facebook knew it would balloon to 10 times its then size. The company launched three projects to find out what platform would best support the surge in user numbers. The teams were to investigate PHP, Quercus and HPHP. A friendly competition was set up, and the teams met regularly. “By definition, two of these three projects would fail,” Schroepfer noted.

It became apparent to the project leaders that HPHP — now known as hiphop — would be the most suitable. Yet Schroepfer said his instinct would have been to pick Quercus. “If I had chosen back then, I would have made the wrong choice,” he said. Another advantage is that hiphop uses 50% less CPU to handle the same traffic.

Move Fast: New engineers are sent on a six-week bootcamp when they start at Facebook. But in their first week at work, they have to make some change to the live site. “The expectation from the very beginning is that you make a contribution,” Schroepfer said.

Another event held in the company is the Hackathon, an overnight immersion in invention. The event typically starts around 7 pm, with a meal at 2 am and runs until breakfast. Schroepfer said he is often asked what can reasonably be expected in such a short time. But the hardest part of a big project can often be the start, he said. The Hackathon kicks them in to high gear very quickly. Outputs “could be a very thin demo or a mostly working prototype,” he said.

And when the projects are presented, the feedback is often the most valuable part. Most go nowhere, but some important innovations such as video, HPHP and Facebook places have come from it, Schroepfer said. However, not all projects are tech related. One team redid an outdoor deck and garden, and Schroepfer joked that the company’s three-year finance plan was probably pulled together in the same fashion.

What Would You Do if You Weren’t Afraid? Schroepfer said this one is fundamental to the success of Facebook. This is not being afraid to take tough decisions even if it hurts in the short term, he explained. An example was to restrict notifications from the site’s 200 million games players feeding in to the Facebook Platform. Schroepfer compared the sheer volume of notifications to the tragedy of the commons. “The user experience became pretty horrid,” he said. Although the company knew developers would not like it, they proceeded with the changes.

Schroepfer said innovation matters to Facebook because the company sees all online activity becoming more social. Another of the company’s sayings is, “Our journey is 1% finished.”  All the time, the firm keeps a close eye on data and what works for users. An example is the ads shown on the site, he said. “Really well-honed personal advertising is more like a recommendation from a friend.” Facebook’s vision is of people liking ads as content.

Asked about competitors, Schroepfer said they saw Google as one, but were also very wary of potential disruptive technologies that could upend their business model. The company will make over $1.2 billion in 2010, and he said the concentration is still on the user experience over monetization at this point.

Schroepfer also pointed to innovation that has had a positive impact environmentally. Hiphop  has cut down on the need for servers, he said. And the two, new, giant data centers in the U.S. were designed to reduce power consumption. Typically, these buildings have servers blowing out hot air, while air conditioners work to cool the environment and prevent the servers overheating. The data centers were built to use outdoor air and cut down on the need for cooling, he said.

Photo of Mike Schroepfer courtesy of J.D. Lasica on Flickr.