It was John, though, who came up with Sergey's favorite line. "The last unbastardized search engine," he wrote. Sergey had a fondness for the word "bastard."
When the mockups of the first ads came back from our freelance artist, the company toasted them with an email flame circle, scorching them from all sides. Most of the comments centered on issues of aesthetics and were easily accommodated with minor changes. Sergey had only one strong objection. The artist had illustrated the line "They really, really like us" with a stock photo showing a stereotypical movie star in dark glasses and head scarf, holding a long cigarette holder. Sergey hated everything about the image, from the artificiality of the obnoxious Hollywood personality to the way the picture promoted smoking.
Larry was even more blunt. "I won't run cigarette ads," he declared.
Feedback trickled in for days. Sergey didn't like a particular shade of green we had used. Cindy thought an ad looked too much like a competitor's. One model was too young. Another too old. I unleashed Sergey's own logic like a sheep dog to contain the flock of ad hoc critics.
"It's unlikely we'll ever come up with fifty ads that all of us like," I pointed out. "And there's no guarantee that if we do, they will be the ads that our users also like. That's why we're testing these."
"Very good point," Sergey agreed. "We don't need everyone to be happy." But still the critiquing continued unabated.
"Too slow."
"Too many stripes."
"Too many words."
After another week of flung arrows, Sergey weighed in again.
"We should try a bunch of these out," he said, "and see how they perform. We're wasting time by not running them and getting the ultimate feedback—clickthrough. Just run them. It's not a TV campaign. Run the hundred banners, see how they do, and then revise them."
I was glad Sergey had spoken. It was time for the banner bus to leave the depot.
That's when Shari, the offline brand manager, threw herself under it. "Sergey," she said, "I agree they don't all have to be perfect, but we risk hurting our brand if we don't set a quality standard. This is why companies hire experienced marketing people. Please ... trust our judgment."
Now I was caught between Shari, whose professional opinion I valued, and Sergey, who was advocating for my right to test ideas in the marketplace. I made the quick changes I agreed with and gave a green light to the rest.
The trades had been negotiated. The creative work was done. All that remained was placing our ads on partners' web pages and seeing what worked best. That's when I learned that every dot-com didn't run like Google. The people our partners put in charge of managing their advertising inventory were intentionally recalcitrant or unintentionally incompetent, or both. One site wouldn't commit to dates for delivering the impressions promised in the contract we had just signed. Another refused to return repeated calls and email messages. Even Netscape couldn't confirm they had actually run any of the millions of ads they were supposed to have delivered for us.
When the ads did run, the results were disappointing. I wasn't expecting much because I had seen clickthrough rates (CTRs) dropping across the web.* Only zero-point-five percent of those viewing online ads had been clicking on them when I left the Merc. I assumed the rate had continued to decline. So when our ads started running, I was skeptical they'd reach the three percent CTR Sergey had set as a goal. They didn't. Most of our banners pulled less than three-tenths of a percent—a disaster in Sergey's eyes. He demanded we stop the ads immediately because he felt we were wasting our inventory. In his view, we should substitute new creative for any ad performing at under one percent once we had shown it five thousand times. Given that we had tens of millions of impressions to use up, that would have meant creating and managing thousands of banner ads, since even successful ads would "burn out" over time as they became overexposed.
Moreover, I didn't trust the numbers our partners gave us. Netscape claimed one ad had a 476 percent CTR, which, being impossible, skewed the metrics for our entire campaign. I asked our own logs-analysis team to verify the traffic our ads actually drove, but no one had time to hand-code our banners to make that possible.
Each day I would gather the reports for every ad we had run on every partner site, plug them into a spreadsheet, and hand-deliver printouts to Larry and Sergey. I'd highlight the best performers and let them know which ads we'd be rotating in or out. The association with actual data seemed to improve my standing in the eyes of our founders. They scanned every cell in the spreadsheet and asked me why certain ads were up or down or performed differently on different sites. I didn't always have the answers, but I could point to the numbers and speculate. I became a convert to the power of data persuasiveness and swore I would make all my future arguments only when I could back them up with real-world metrics.
With attention now focused on the ads' performance instead of my own, the pressure from above eased somewhat. We had created our first hundred ads quickly and cheaply and the production costs were going down. Our most effective ads featured nothing more than a white background, a search box, a logo, and some text ("The answer's in here"; "Who are you looking for?"), and webmaster Karen and Wacom† wonder Dennis Hwang could crank those out in fifteen minutes. I had actually accomplished something to justify my existence in the eyes of our engineering overlords.
The shelves in my cupboard of confidence were no longer empty but for crumbs and cobwebs. Still, each night as I tucked my ego tight behind shuttered lids, I could just make out the sounds of a grindstone rubbing against a metal blade, slow muffled footsteps, and the whistle of an ax falling toward a wooden block.
It kept me on my toes.
Chapter 7
A Healthy Appetite for Insecurity
WHAT DID IT feel like—the experience of coming to work at Google when it was fewer than sixty people? Let me give you a few impressions. Before I started at Google, I had never said any of the following on the job:
"Yes, I see the eight shelves of programming books. Where do we keep the dictionaries? No, I can't just print out the words as I look them up online."
"Is it a good idea to have all those bikes leaning against the fire door?"
"Sorry. I was aiming for Salar. Did I get the printer? Super soakers are really inaccurate at more than five feet."
"Who do I ask if I have questions about Windows? No one? Really?"
"Wow, Larry. Who trashed your office? Well, it's just that ... uh, never mind."
"Wouldn't it be easier to buy rollerblade wheels that are already assembled?"
"Is there any way to set the sauna for more than half an hour?"
"Is it okay to go into the women's locker room to steal some towels?"
"Oh, sorry. Didn't realize anyone was napping in here."
"See, you knock down more garbage cans if you bounce the ball instead of just rolling it straight at them."
"It's in the area behind the coffee-can pyramid, right across from where the Big Wheel is usually parked."
"I tried to book ninety minutes, but the schedule was full. So I only got an hour. Could you focus on legs and feet? I think I pulled something running this morning."
Insecure in the Knowledge Your Contribution Matters
I needed to stretch. I'd been staring at my screen for two hours thinking up new banner ads, responding to users, and working on an "email-a-friend" program that Sergey thought had the potential to go viral.
"Your user name is not valid," I wrote for one of the program's error messages. "It may have a bad character. That's not a reflection on you." I was getting a little pixel punch-drunk and it was affecting my judgment.
I left my cubicle in the marketing pod and meandered off in search of glucose and caffeine. Google was growing. The company was still contained in a single building when the millennium began, but the offices lining the outer edges of the Googleplex had all been occupied.
One day a crew of Samoans, their thick biceps shrink-wrapped in coconut-leaf tattoos, arrive
d to fill the open space with cubicles. The area was now partitioned by a maze of cheaply acquired, mismatched fabric panels, the flotsam and jetsam of the dot-coms that had suddenly started sinking all around us. Fast-food toys, manipulative puzzles, empty soda cans, and geek-chic objets d'Nerf feathered the work nests. Ratty couches shambled through open areas and settled on brightly colored crop circles cut into the carpeting, offering lumpy, coffee-stained comfort and filling in for laundry hampers. I brought in a couple of four-foot-high inflatable dinosaurs and left them to graze on the new flooring.
Walking the gray-padded arroyos, I glimpsed many heads silhouetted by code-filled screens. It may sound deadly dull, but there was an energy to the place—conveyed in quiet conversations, snatches of laughter, the squeak of dry-erase markers on rolling whiteboards, exercise balls bouncing, and electric scooters humming down hallways.
Yoshka ambled past, ears flapping, collar jingling. Someone flopped on a couch, took off his skates, and dropped them on the floor. Someone ground coffee beans for an afternoon espresso. A pool cue slapped a cue ball. It passed the aggression on, smacking an eight ball loitering in its path and sending it into the deep funk of a faux leather pocket.
I sensed the tension of potential—building and bound only by time—like the feeling of crossing the tracks in front of an idling train. Great efforts were being made, and the energy they required rippled outward seeking physical release.
Sometimes that physical release took an intimate form behind closed doors with a willing partner.
We had a crash cot in a windowless nap room for those who had reached the edge of endurable fatigue and lurched beyond it. One afternoon a staffer peeked in and found two engineers on the bed, engaged in an act of noncomputational parallel processing. It was decided that the space—once sanitized—could be put to better use as an office. No punishments were administered, no stern policy reminders sent out. Those who might have cast stones couldn't find adequate purchase on the moral high ground, and so unofficial UI experimentation continued, just later at night and relocated to offices lit only by passion and the glow of multiple monitor screens. "Hormones were flying and not everyone remembered to lock their doors," recalls HR manager Heather Carnes.
Larry and Sergey encouraged everyone to channel their excess energy into roller hockey instead. Any employee who signed up was issued a free NHL jersey emblazoned with his or her name and Google's logo. Hockey provided yet another metric by which Googlers could be evaluated.
"There is no better way to get to know someone," George Salah, a regular participant, believed. "To have their true colors come out, play sports with them. You get to see how aggressive they are, if they're ruthless or not, if they're capable of giving a hundred and ten percent." As a result, no one held back when fighting the founders for the puck. In fact, the harder you played, the more respect you earned. It was not uncommon to see blood and bruises when the games ended.
I never strapped on skates and joined Google's Thursday parking-lot hockey game, but I wasn't immune to the competitive intensity driving the company. I found my outlet in the rowing machine shoehorned into a corner of the rec room that engineer Ray Sidney had cobbled together. I'd drop into the gym between meetings, sit on the sliding seat, strap my sandaled feet into the footrests, and breathe deeply. I'd reflect on the electronic cholesterol clogging my inbox, the uninvited additions to my work queue, or some viewpoint variance I'd had with a colleague. I'd grip the padded pull bar, close my eyes, and jerk with all my might, sending my stationary craft slicing outward to placid waters far from the source of my current distress.
It wasn't approved form, but I wasn't trying to win a regatta. My goal was to generate the maximum number of ergs in the minimum amount of time, to best the score posted on the "Google Rowing Club" whiteboard resting against the wall. Claiming the title "best" in any category at Google took on added significance given the capabilities of those with whom we worked.
At age forty-one, I had much to prove. Sometimes it took me a little while to parse a TLA everyone else seemed to know.* I voiced objections based on "irrelevant" experience gained over a twenty-year career. I drove a station wagon that smelled like baby wipes and spit-up. I didn't want the unripened grads with whom I shared the locker room to assume I was slowing down physically or mentally.
One afternoon the receptionist called me.
"Doug, can you come downstairs? Sergey asked if you could load his scuba gear in the car. He said you're the strongest guy here."
"Sure," I replied. Halfway to the lobby, I slowed down. Then I stopped. The founder of the company wanted me to do his scutwork. That couldn't be good, could it? But Sergey felt I was uniquely qualified. That was a plus, right? I was glad to be singled out, but embarrassed about the reason for it. Was I that in need of recognition?
Google's obsession with metrics was forcing me to take stock of my own capabilities. What did I bring to the table? What were my limits? How did I compare? Insecurity was a game all Googlers could play, especially about intellectual inferiority. Everyone but a handful felt they were bringing down the curve. I began to realize how closely self-doubt was linked to ambition and how adeptly Google leveraged the latter to inflate the former—urging us to pull ever harder to advance not just ourselves but the company as a whole.
Toward the end of my Google run, a newly hired senior manager put into words what I had discovered long before. "Let's face it, Doug," he confided, "Google hires really bright, insecure people and then applies sufficient pressure that no matter how hard they work, they're never able to consider themselves successful. Look at all the kids in my group who work absurd hours and still feel they're not keeping up with everyone else."
I had to agree that fear of inadequacy was a useful lever for prying the last erg of productivity out of dedicated employees. Everyone wanted to prove they belonged among the elite club of Google contributors. The manager who articulated that theory, though, considered himself too secure to play that game. Which may be why he lasted less than a year at Google.
Keeping It Clean
Google's embrace of "organized chaos" extended to our workspace, which might charitably have been called a mess, or less charitably, a pigsty.
The locker room had come with showers and saunas. The odor we added ourselves: a pungent mix of soiled jerseys, scuffed knee pads, muddy pucks and headless hockey sticks, grip tape adhesive, deodorant dispensers, ripped underwear, and expired aftershave. Google soon provided towel service and installed low-energy Swedish washing machines that took a week to spin through a single cycle—introducing a ripe finishing note of undone laundry, abandoned athletic socks, and mildewed terry cloth. Imagine a geek fraternity claiming squatters' rights in an insurance office.
The one area in which hygiene was fastidiously maintained was engineering. Not the engineers' physical space—they were apparently all feral children—but their operating principles.
Urs insisted on adopting the best practices he had seen in more industrial settings to things like source control and compiler warnings. "We'd make sure the compiler actually failed if it had a warning, so you couldn't ignore it," he told me. He formalized the most important elements into a style guide, which became a mandate enforced by Craig Silverstein.
"I had no desire for a style guide," Craig noted, "but Urs was really insistent."
The biggest question was which programming language to use. Craig wanted to use C. Urs preferred C++. Urs prevailed, but he agreed to restrict Google coders "from using the bad parts of C++."
"What are the bad parts of C++?" I once asked Craig.
"Most of it," he said with a straight face.
Craig believed Google would need an integrating force to prevent redundant or conflicting efforts, maintain standards, help set priorities, and provide feedback during performance reviews. He made it his goal to be that force, to be "the one who knew what was going on everywhere."
"Until we were about a hundred people," he recalls, "I'd go
around and talk to everyone. I'd say, 'How's your work going? Do you need any help?' Some people got really upset about that. 'This guy keeps bothering me. What's he doing?' Urs had to take me aside and tell me to stop." Craig realized that "they didn't need someone to pay that close attention. Everyone was still just one degree away. They knew who to talk to. There was very good communication."
The one thing that did need a high level of attention was the code itself. To look for potential problems, Craig started scanning every automatic alert generated whenever someone checked in a codebase change—no matter how minor. But a lone proofreader couldn't keep up with the growth of engineering's output. So Urs instituted a formal code-review process.
"You get to pick one good engineering practice to be your cultural touchstone," Craig said, "and for us it was code reviews." To initiate a review, a coder would send out a pointer to an online design document. Anyone could respond with comments, but an official reviewer had to sign off on the actual code.
The benefit was obvious. "Finding problems in the beginning," engineer Ron Dolin explained, "was a hundred times more efficient than finding them later on."
Still, as Google grew, not all the programmers subscribed to the idea that their code needed proofing or that it was their responsibility to look over other people's work. As Craig explained it, "We put a process in place to prevent submitting code without review, unless you lied."
To get around the process, people would perform cursory reviews. "I'd send out a big, giant piece of code," Craig said, "and they'd send back, 'Looks good.' I suspect there's more that could have been said about that code."
When new employees started, Craig reviewed their code himself, inculcating them into the system of collaborative authoring. Some people hated it. "paul Bucheit hated it," recalls Craig. "Noam* hated it. But Paul ended up doing a lot of code reviews on his own and being really supportive of that methodology. Noam thought it was a waste of time—he's like Larry and Sergey, a research developer. He'd say, 'I'm spending more time reviewing this code than I spent writing it—how is that a good use of anyone's time?'"
I'm Feeling Lucky: The Confessions of Google Employee Number 59 Page 10