Much has been written about Google's free meal plan (one estimate put the cost at seventy-two million dollars per year),* but the basics of the program were simple: lunch and dinner were free, and we could eat as much as we liked from our first day with the company until our last. Like most Googlers, I spent less than half an hour at lunch and, if on deadline, would just retreat with a plate to my desk. Without the café, I would have lost twenty minutes getting to a restaurant, half an hour eating, and another twenty minutes getting back. I would have stopped thinking about Google as soon as I cleared the front door so I could focus on consuming fatty, salt-saturated foods on my way to increased sick days and a premature death. Looked at that way, the policy made sense to me.
Charlie did have a budget, however, and given his complaints about its limitations, everyone was shocked the day he served lobster for lunch.
"Don't worry," Charlie assured the crowd, "I got a special deal because they only had one claw." No one dared ask where he had shopped for one-clawed lobsters. Perhaps they got in too deep with loan sharks. They tasted great. That was all we had to know.
Besides, Charlie made sure nothing went to waste. "In the early days, Charlie used to call me down to eat all the leftovers," engineer Chad Lester fondly recollects. "At the very beginning I could sometimes do it. I'd eat my normal lunch and wait for everyone else to eat what they wanted. Then around two-thirty, I'd be hungry again. If I didn't show, Charlie would come get me, and I would eat everything that remained."
"Once I ate an extra nine pork chops," Chad recalled with a happy smile. "Another time Charlie gave me a mixing bowl full of ice cream, whipped cream, nuts, candy, and chocolate sauce. It must have weighed a couple pounds. That was insane."
Great food also had the ability to attract great talent.
"I don't know what to do," senior engineer Luiz Barroso moaned to Jeff Dean the night he had to decide whether to join VMWare or Google. "I've made these lists. I've assigned points to all the pros and cons, and it's tied at 112 to 112."
Jeff knew that the day of Luiz's interview at Google, Charlie had served crème brûlée for lunch. "Did you factor in the crème brûlée?" he asked. "Because I know you really like crème brûlée."
"Oh no! I didn't consider that," Luiz admitted. The next morning he accepted Google's offer.
It wasn't just a nice gesture to treat job candidates to a decent meal, it was part necessity and part test of temperament. Since interviews stretched into daylong affairs, it was important to give applicants enough nutrition to sustain them. We could easily spot the aspirants—they'd be sitting out on the deck sweating in navy blue suits while all around them Googlers in shorts and sandals chatted and chewed. Any Googler who happened to be within earshot could pepper candidates with questions, and the answers could influence a hiring decision as much as anything in the formal interview process. Giving off a "Googley" vibe mattered.
I never knew whom I might bump into while waiting for a fresh platter of polenta to be put out. At first, celebrity drop-ins tended to be tech luminaries like pundit Esther Dyson, Sun superstar Kim Polese, or the chairman of Intel, but as Google's fame grew, you were just as likely to run into Nobel laureates and internationally known politicians, people like Muhammad Yunus, Queen Noor, Bill Clinton, or Jimmy Carter, pushing trays along the aluminum rails under Charlie's watchful eye as the Grateful Dead wailed from wall-mounted speakers. Journalists from Japan and France stood in the middle of the café pointing cameras and murmuring in their native tongues as print reporters from Time and Fortune and BusinessWeek huddled with Larry or Sergey to chew on chicken sandwiches and ruminate on the future of search. The café took on a circus quality and lost some of the intimacy of the first few months. But the food always brought me back.
Blending equal parts fanaticism, ego, and artistic temperament, Charlie served up a mélange of exotic tastes mixed with intelligent discourse and the fellowship of shared interests. He was a blur in the kitchen, throwing inadequate resources at impossible demands, with sweat beading his brow and food stains augmenting the all-over tie-dye of his custom-made apron. The lesson of the data center applied to the kitchen as well: cheap production units pushed to their limits offered superior performance. Individual servers, whether of web pages or of steamed broccoli, might give out, but the system wasn't truly broken as long as it kept delivering results. To their undying credit, Charlie, Jim, and the rest of the Google kitchen crew never experienced a catastrophic failure. Day after day after day, they fed us—their infrastructure running on elbow grease, ingenuity, and heart. It was a very Googley way to be.
Chapter 8
Cheap Bastards Who Can't Take a Joke
HOW MANY GOOGLERS does it take to screw in a light bulb?" Sergey asked the UI team in late February 2000. His complaint was about browser buttons—a trivial bit of code that allowed users to add Google search links to their web-surfing software. We were debating names and design details by email, and the list of people involved had grown to ten, including Urs and Cindy. Sergey found that ridiculous.
"If we continue to do all of our work at this kind of pace," he observed, "I find it highly unlikely we will meet any significant number of our goals." Instead of exhaustive user testing and internal deliberations, he argued, "We should just put up the service and test it when we get a chance. Don't forget, we can change it at the drop of a hat—or we should be able to."
Sergey's perspective on "launch first, iterate later," was nothing if not consistent. "I don't think we should have any meetings about a project like this," he said, "or any group emails except the one to users announcing the launch. Having everyone involved in every issue is not a good use of anyone's time."
Sergey wanted one person to take charge: Salar. As product manager (PM), Salar could talk to Karen to set up the web page, to Marissa or Bay to check on the UI, and to me for branding, then work with an engineer to make changes if necessary. Salar had no training as a PM—no ivy-covered MBA, no internship at an overpriced consulting firm, no career spent hauling himself one rung at a time up a corporate ladder. But Sergey thought he was bright enough to figure it out. Sergey felt the same way about Susan, who soon left marketing to take on a role similar to Salar's. More important, Larry and Sergey trusted them not to get in the way of what the engineers wanted to do.
Larry officially anointed himself chief of products in March 2000, taking charge of the entity coalescing around the kernel of product management. He would oversee any project requiring engineering resources—the "Googlettes" hatching all around the building—including the directory and WAP (Wireless Application Protocol) search for cell phones. Larry decreed a meeting be established at which views would be heard from all corners of the Plex, disagreements would be aired, and edicts would be issued. He dubbed it "product review." Google had birthed a process.
Product review met in Larry and Sergey's office. I arrived early to get a seat on the black pleather couch. Otherwise, I'd have had to balance my laptop while sitting on a three-foot rubber ball. A large metal exoskeleton—the prototype for Larry's book-scanning project-held a camera and an array of lights pointing down at the coffee table in front of me. Karen White, Marissa Mayer, Jen McGrath from the front-end team, and Craig Silverstein worked around it, connecting cables to a projector so we could display mockups against the office wall.
Sergey leaned back in his desk chair across from us, reading and eating a sandwich. It was hard to tell if he was paying attention.
Marissa set the agenda, determining which products to discuss and who should be at the meeting to present the case for or against proposed changes. Marissa was twenty-four. A slim, blond Wisconsinite, she was well versed in UI issues and she never met a problem she didn't try to fix—immediately. Her mind worked so quickly that her buffer overflowed, filling all available conversational space with a flurry of words. She would download what she was thinking, punctuate it with an ellipsis laugh, and then preemptively address all objections or alternative viewp
oints that could conceivably be expressed. It took me a while to get used to, even though I had dwelt among New Yorkers.
Marissa took the role of product review gatekeeper as seriously as an embassy guard in a hostile nation. She ardently argued her views about the best way to help Google prosper while protecting the inner sanctum from antagonistic ideas. Over the months to come, as she began casting a larger shadow on both the product-development process and Larry's personal life, she would iris down the exposure others had to the chief of products. We were, of course, free to talk with Larry outside product review, but since the meeting was now the place where decisions were officially made, those conversations could easily be viewed as redundant and a waste of Larry's time.
No one wanted to waste Larry's time.
Larry himself remained unassuming about his new role. He sat at his desk at the far end of the cramped office, glancing up from his array of computer monitors when we presented something of interest to him. "How does Yahoo handle that?" he might ask, sending us on a round-the-web sightseeing tour to learn if someone else had already solved our particular problem. When we presented competing views, Larry would call for the data supporting each side, then offer his Solomonic verdict.
And when Larry was done, he was done. "I don't want to talk about this anymore. It's not worth discussing. Just do it."
Larry (Sergey, too) ended discussions abruptly when he saw time being wasted on something he had already decided. He hated rehashing arguments and bringing irrelevant people up to speed. Mostly, he couldn't abide having to explain the obvious. Sometimes all it took was a look. Project manager Deb Kelly proposed a new method for putting content online and ran it by Larry. "He did the eyebrow thing," she later recalled, arching one eyebrow like Spock. "And he didn't even give me a reason. He was just basically—'No.' It was clear that it was a stupid idea."
It didn't take long for me to have my own first brush with Larry's debate-killing decisiveness. Larry wanted to relabel the "cache" link on our results pages to read "show matches."* Clicking the link would still take users to a cached version of the page, but it would now highlight words on that page matching the user's query. I found the name change confusing and felt sure our users would too. It was a wording issue. That was my domain. I voiced my objections at the UI meeting and won agreement that we wouldn't make the change without at least testing it first.
The next morning I woke to find "show matches" had replaced "cache" on all our results pages.
"The UI team does not have control over these decisions," Bay Chang reminded me when I asked what had happened. "Larry wanted 'show matches.'" Larry had stopped Marissa as she headed home, and together they had decided to implement the edit.
"I carefully considered all the feedback," Larry said when I confronted him. "And I don't want this to be discussed endlessly. There are a lot of more important things for us to do. We've already repeated all the arguments, which means it's time for a decision."
I had thought that decision time had passed during our UI team meeting.
"Why did Google stop showing cached pages?" asked the first user complaint we received.
"I loved the cached pages! Why are they gone?" demanded a dozen others. It wasn't the End of Days, but given how few people actually used the feature, it qualified as a minor plague. Most alarming, reporters complained to Cindy and passed along notes they were getting from their readers. A week later we changed the label back to "cached" and I plotted three new data points on my Google graph:
Nothing was final until Larry said it was.
Larry communicated directly to the people who could implement his decisions.
Larry erased what he had etched in stone if the walls crumbled around him.
The third point was the most important lesson to me, because undoing things done wrong would be the crux of my own Jericho moments at Google.
The first of those moments started as a joke.
Are You Kidding Me?
It was the middle of March. Time to put away the space heater and break out the cargo shorts and sandals. April was right around the corner. My first April at Google. My first April first. My first opportunity to undertake the torqued brain aerobics and flop-sweat composition that I came to know as the Google April Fools' joke.
April Fools' Day would become a perennial black hole in my calendar, a gravity well into which my attention would be sucked from increasingly great distances in time. Sergey, on the other hand, loved April Fools'. His sense of humor didn't stop at the boundaries of good taste, and when it came to April Fools', he dynamited decorum and put moderation to the torch. The cruelest month, indeed.
I was headed into Charlie's Café when I ran into Sergey. The day was warm and my mood was full of springtime. Charlie had made my favorite apple galette for dessert. As I pushed my tray through the line I chatted with Sergey about the possibility of doing something fun for April Fools'. He lit up and encouraged me to go for it. That afternoon I sent him a press release announcing that Google had patented an "alpha-numeric system" that eliminated the letters Q, Z, and C. They were redundant and increased page-load times. Sergey granted it was amusing but, he informed me, unnecessary. Susan would head up our April Fools' effort. If I had other ideas, I should coordinate them with her.
Ouch. That hurt. Funny stuff on the homepage felt awfully like online branding to me. I went back to the drawing board.
"Let's see. The message should be positive. Improbable, but not obviously impossible. Hmm. What if Google were so good it delivered results before you even searched?"
Thirty seconds later, I was typing a description of "Ante-Temporal search," a breakthrough development that anticipated user requests. The tone was heavily geekish, but Susan liked it. Sergey thought it was overdone (akin to the pope saying you're overly religious), but encouragingly, he said it had potential. "Maybe you could make it more like this site I saw about kitty porn?" he asked.
Sergey asked Susan to include engineer and bodily-function satirist Ray Sidney in our discussions to make sure the joke ended up with plenty of funny and sufficient tech cred. I dragged Ray and Paul Bucheit into a conference room to brainfart jokes. In that redolent atmosphere, Ray pinned the name "MentalPlex" on our new mind-reading technology. That settled, I hashed out the text for a link on the homepage, a dozen error messages that would be displayed randomly if someone entered a query in the MentalPlex search box, and an FAQ explaining how to visualize a search so MentalPlex could detect it.
Q: I am unable to visualize clicking and have to use my finger to activate the mouse.
A: Click visualization takes practice. Try pushing the mouse button with your eyebrow, then gradually increase the distance between your eye and the mouse.
Marissa pointed out that with our newly launched foreign-language interfaces, we could extend the joke beyond English. "One of the error messages should say that MentalPlex has detected foreign thoughts," she suggested, "and then we can translate the interface text on the results page into German."
In what would become an April Fools' tradition, our webmaster Karen White worked at warp speed right up to the deadline to get the final images and copy in place. We wanted maximum exposure in all time zones, so we would send MentalPlex out at eight p.m. on March 31 and keep it on the site throughout the following day.
Karen pushed the last of the files at 7:55 on Friday night. Soon Google users would see a spinning cartoon spiral on our homepage inviting them to try MentalPlex. I was relieved we'd made the deadline. I was also terrified.
Humor is subjective. Done poorly, it just comes across as pathetic. I didn't want my first attempt at instilling personality into our brand to be branded lame by the digerati, but I also didn't want it to be so much of an in-joke that only geeks would get it. It would be bad enough to tell a joke to millions of people and have no one laugh, but if MentalPlex bombed, I was sure Sergey would just hand all future fun stuff to engineers, condemning me to write product spec sheets and error messages f
or all eternity.
As soon as Karen hit Submit on the push, I started compulsively refreshing my inbox to check for feedback from users. I had a sense that my entire career was resting on whether or not I had nailed the punch line for this one stupid joke, which felt weaker and more juvenile by the minute.
At 8:01 the first email arrived. "Google is great!" was the header. The user liked the way we played with our homepage. More emails started trickling in. People were surprised. They didn't think search engines had a sense of humor. They liked it. They were ☺. LOL. ROFL. ROFLMAO. They really, really liked it. I let the glow of their adoration wash over me from the screen of my laptop. There were a few confused souls who didn't get it at all, and some who felt we should forget comedy and stick to searching, and a handful who threatened to leave if we put graphics on our homepage permanently. One guy who said he was "incipiently epileptic" believed our spinning spiral graphic could induce seizures. But these sour notes were drowned by the chorus of hosannas sounding steadily over the next few hours. It was a love fest of praise and joy and Lo! Was that inaudible rhythm the music of the spheres thrumming softly in the heavens?
I'm Feeling Lucky: The Confessions of Google Employee Number 59 Page 12