I'm Feeling Lucky (57 page)

I always laughed. If it had been real, we would never have let them see it. Especially as, by the beginning of 2004, we were on the road to becoming a public company. Our obsession with secrecy had always been a shared cultural value. We didn't talk to anyone but other Googlers about what we did behind closed doors. Eventually we didn't even talk to them. When special key cards were suddenly required to access one engineering floor, the employees who worked there joked that they were working in the nude. By 2004, access to all information for all Googlers was no longer the norm.

Even MOMA, our free-for-all company intranet, began to show signs of becoming buttoned down. Our engineers had always obsessively collected every scrap of intelligence they could about what was happening within our servers. They analyzed it, kneaded it, and baked it into tasty little tidbits set out on MOMA where any Googler could consume them. The data was anonymized—you couldn't see individual queries or IP addresses, for example-but the number of searches, countries from which they originated, most popular search terms, and other key stats could all be viewed.

MOMA's homepage was originally dense and messy and full of numbers. At the center sat a large graph with colored lines labeled with the names of Muppet characters. The graph represented results quality across different search engines, and the top line, labeled "the Great Gonzo," belonged to Google. When another line veered close to ours, clarion calls could be heard above the gnashing of teeth, ordering that all energies be focused on improving the relevance of our results. Larry and Sergey never forgot that the quality of our search drove our success and never took for granted that our lead was insurmountable. Ironically, the lack of a good way to search MOMA made it hard to use at first, though Google finally hooked up one of its own search appliances to fix that problem.

The most useful aspect of MOMA for me was the phone list, which contained the title, email address, and photo of everyone on the payroll. My picture was there. Sort of. My original photo captured more reflected flash than facial features, so I swapped it out for a press photo of Deputy Director Skinner from
The X-Files.
The resemblance was eerie, and his picture conveyed the gravitas and focus my own photo lacked. Other MOMA photos showed samurai warriors and masked figures with titles like "Shadow Ops" and "Black Ops." Yoshka, Urs's Leonberger, was listed as "Google's top dog." New Googlers looking me up for the first time would inevitably email me, asking about my uncanny resemblance to Mulder's boss. I'd assure them that the truth was out there.

In fact, the truth was on MOMA. I came to assume that any information I needed about Google could be found on the intranet, from the status of products in development to the number of employees at any point in the company's history. It was a shared wellspring of data that all Googlers could tap to test hypotheses, build prototypes, and win arguments.

In mid-2003, Susan put some product plans and strategic documents on MOMA that required a password to access. She was concerned that the sales team might accidentally spill too much to clients. As head of product management, Jonathan told her to make the documents accessible because Google so strongly valued the free flow of information among staff members. Only performance appraisals and compensation were off limits. "This is extremely unusual for a company to do," Eric Schmidt often reminded us at our weekly TGIF meetings, "but we will continue trusting everyone with sensitive information unless it becomes a problem."

In September 2003, it became a problem. Information about our revenue numbers and Larry and Sergey's stock holdings started showing up in news reports. Eric immediately clamped down, telling Omid and me to stop including revenue numbers in TGIF presentations. Passwords on MOMA were no longer forbidden. It was a shame, Eric observed, that reality had finally come to Google.

The source for the stories turned out to be a low-level administrator feeding information to an outsider. She was asked to leave. In January 2004, though, long after that first small leak had been plugged, a much bigger crack appeared in our wall of secrecy. The same month we hired our first corporate security manager, John Markoff from the
New York Times
wrote a series of articles in which he reported details of products in development and the results of an internal audit conducted in preparation for a possible IPO. The information had been extremely confidential and closely held. The leak was ultimately traced to a senior manager who had known Markoff for years. He left the company as well, though the true reason for his departure was not made public, leading to much speculation.

From that point on, I had to ask for access to the project information I needed to do my job. It felt odd, as if with each ironclad, password-protected gateway the company installed, it locked out a little more of its original corporate culture.

Shortly before going public, Google clamped down completely. According to SEC rules, every employee who had access to intimate knowledge about the state of the business would be restricted from freely buying and selling the company's stock. I, and most others, gladly traded ignorance about our bottom line for the bliss of being able to cash out whenever we were ready to do so. The days of innocence in the garden of data had officially come to an end.

In February 2004, Yahoo dropped Google and began using their own Inktomi-based search results instead. We barely noticed. We stretched in the skin of our new headquarters and settled in to a new level of hyper-productivity. Everything needed to be done
right now
and everything was
very important.
New people were climbing onboard every week and taking control of projects in motion.

Cindy kept an eye out for any signs the news cycle was turning against us. She urged us not to let cracks appear in the shiny gold sphere of Google's public image, and every few months she sent reminders to all Googlers that when the press came calling, the calls should be forwarded to PR. "There is no such thing as, 'off the record,'" she cautioned us, because "reporters are fiercely competitive and will tell you whatever you want to hear just to get the story." The last thing she needed was a very public slip on something as important as a new product launch. But sometimes things go wrong.

Engineer Orkut Buyukkokten came to Google in the summer of 2002 from Stanford, where he had become intrigued by the idea of social networks—a way to connect with friends and acquaintances online. As a student, Orkut had written a networking program for his classmates called "Club Nexus," and once he settled in at Google, he requested to spend his twenty-percent time coming up with an improved version. It was December 2003 before his project, code-named "Eden" and later renamed "orkut,"
^*
was ready to be tested with a broader audience than just Googlers. That's when the fun began.

Orkut built his eponymous service entirely on his own. It was a prototype to gather data, to try things out, to experiment. He wrote the code, designed the user interface, set up the databases. He didn't intend for it to be a full-fledged Google product, so to accelerate the development, he used tools that were commonly available outside Google. They came from Microsoft. The server running orkut wasn't even located in a Google data center, but at the home of the weather site
Wunderground.com
. Orkut knew his system would never support Google-sized audiences, but it should safely scale to handle two or three hundred thousand users. Membership in orkut would be by invitation only, so he would be able to throttle growth by controlling the number of invitations the system distributed.

Marissa was the consumer product manager. She saw orkut as a small startup within Google, operating autonomously to prove that a single engineer with a new idea could build and test a product without enduring the delays of Google's increasingly bureaucratic development process. Larry and Sergey encouraged her to manage orkut as if it were an independent operation.

Other prototype projects by individual engineers were available in an area of our website called Google labs, but Marissa didn't believe orkut belonged there. It would be an exception—a standalone product, without Google branding, launched with lightning speed. If there were problems, few people would notice and there would be time to fix them. If anyone asked whether orkut was a Google initiative, we would answer that we could "neither confirm nor deny" a relationship between orkut and Google.

That did not sit well with Cindy, who had no desire to play coy with the press contacts she had so carefully cultivated over the years. "Reporters are not stupid," she warned the executive staff, "and we'll look silly saying this." All it took was one online search to find Orkut's connection to Google, and once the press had that, orkut would be branded a Google product whether we denied it or not. I worked on messaging to make it clear that orkut was developed by a Google engineer but was not an official Google project. Marissa rejected any such compromise. She was adamant that no explicit Google connection be revealed on the orkut site itself.

Sergey stepped in to resolve our standoff shortly before he, Larry, and Eric headed off to the World Economic Forum in Davos, Switzerland. "Let's make it an experiment," he said. "We'll launch without Google affiliation and neither confirm nor deny a relationship. If it gets out of control, then contact me in Davos and we'll come up with a new plan. Any problems with that approach?"

Oh yeah. Cindy had a big problem with that approach, and she let Sergey know it. This experiment, she pointed out, could destroy our brand reputation as well as her professional credibility. Sergey retreated to a fallback position. He endorsed launching orkut without any Google branding, but he conceded that if asked, we could admit orkut had been developed at Google.

On January 21, 2004, the day before the scheduled launch, Jonathan voiced reservations. He advised Eric Schmidt that we should wait to launch orkut until after our global sales conference and the company ski trip, a two-week delay. Cindy told Jonathan his effort was appreciated, but it was too little, too late. We had wasted weeks trying to live with the conditions Marissa had set while repeatedly advising that it was wrong to lie to the press. Product management of orkut had been bungled from day one.

Urs, our Google Fellow, raised his own questions about the timing. He had been hearing all day from engineers about unresolved issues: that orkut was running on a single machine with no easy way to scale, that there had been no proper load testing, no security review, and no agreement on the privacy policy. Clearly orkut would not be able to handle the influx of traffic once word got out to the geek news site Slashdot,
^*
which would take about fifteen minutes. It would be smarter, he said, to clean things up for a few days, wait for the execs to return from Switzerland, and avoid a huge mistake.

We all breathed a sigh of relief when Jonathan confirmed that he had spoken with most of the executive team and they had agreed with Urs it was better to delay the launch. It was a commendable idea to accelerate the launch process, but just too risky given all the red flags that had been raised. We would begin moving orkut off its
Microsoft.NET
server to the standard Google technology platform so it could scale more easily and then, most likely, launch it officially on Google labs. I had proposed exactly that strategy and believed it would set user expectations appropriately—everything on labs was by definition an experiment, subject to drastic changes or sudden shutdowns. Marissa thought a launch on labs would irritate users, because they would be able to see orkut, but not to try it unless they were invited by someone who was already a member. That didn't worry me as much as users thinking, as with Froogle, that a half-baked orkut was Google's flagship product in an entirely new online sector.

I kept working on language for orkut's interface pages, user notifications, and community descriptions, so I was ready when, a day after Jonathan's reassuring news, Sergey sent word from Davos that we would, in fact, launch orkut that afternoon. Sergey had made a commitment to support a rapid implementation and he was standing by it. We would compromise by putting on
orkut.com
a small tag saying, "In affiliation with Google."

An hour before launch I sent Orkut text for his homepage: "orkut is an online community that connects people through a network of trusted friends. Join orkut to expand the circumference of your social circle." I included a disclaimer: "Great relationships begin with honesty, so we'd like to let you know up front that orkut is still a beta service. That means that it may appear a bit flaky at first. It may even require quiet time alone to work out some issues. We hope you'll bear with orkut as it strives to better itself. After all, that's what good friends do." At one in the afternoon, the first twelve thousand invitations to join orkut started going out via email.

In the social networks Orkut had built at Stanford, the users had been polite, respectful, and courteous. The users of
orkut.com
, however, were not. They immediately began looking for ways to break the system and to fill it with porn and spam. They found them. It was possible to search for every user in the system and then send them all email with hundred-megabyte attachments or write a script to add everyone as a friend. Orkut the project's creator had never seen problems like those before. He had to take the site offline almost immediately to fix them. I wrote an error message saying we were making improvements. It was to be expected given that orkut was just an experiment.

Google engineers didn't accept that excuse. They peeled back the skin of orkut's architecture and picked every bone they could find. Why hadn't orkut received a full security review? What was driving the rush to push it out the door? Wasn't it evil to place our need to launch before the security of our users? Why had we even tried to conceal Google's involvement? The questioners piled on. Marissa gamely defended the decision to move ahead by describing a startup-within-a-startup mindset that meant taking risks and patching things up as you went along. If decisions had been made differently, she indicated, and the Google name not attached to orkut at the last minute, everything would have been fine.