Input from outside the Googleplex? That didn't carry the same cachet. As I saw with the April Fools' feedback, we discounted user dissatisfaction unless it could be clearly demonstrated to cause significant shifts in actual behavior. Our arrogance ultimately became a nasty undertone in conversations about Google taking place in the press and among those trying to do business with us, but I rarely saw it expressed by Googlers toward their own colleagues.
When Googlers did engage in blatant opinioneering, they did it on Googlers-MISC,* the Las Vegas of mailing lists, where nothing was out of bounds. On MISC, Googlers offered theories about proving P=NP and the best way to levitate frogs. They debated the brand of bottled water Charlie should stock in the micro-kitchens, a discussion encompassing total dissolved solids, bottle size, and the health benefits of naturally occurring uranium. After a week of this, Charlie's patience wore thin enough that he threatened to pull all the bottles and let the staff drink pond water.
I started a rancorous MISC thread by suggesting we eliminate the free donuts Ed Karrels delivered on Fridays in favor of fresh bagels. That one bounced around for years. A request with the subject line "Your mother doesn't work here," asking that Googlers do more to clean up after themselves, set off an explosion of self-righteous rage about the value of engineering time, the role of women in the postfeminist workplace, and the second floor's desperate, heartbreaking, and absolute need for more flatware. That lasted almost as long as the discussion about the corrupted physics in Star Wars: The Phantom Menace.
MISC was where I went to take the pulse of Google's culture, and by all signs, that culture was vibrant, diverse, and occasionally obsessive.
The Perpetual "Why?"
The mental image of engineers I carried into the Googleplex was one of introverted nerds with retarded social skills and skin that never experienced direct exposure to sunlight. I anticipated they would be easily cowed by strong personalities with loud voices and authoritative manners. Not that I would necessarily fall into that category, but after seven years working with acid-tongued inquisitorial journalists arguing points ad absurdum, I expected things at Google to be easier. I thought the most forceful pushback would come from the springs in my keyboard as I typed directives into my computer.
I discovered that the hardest-bitten investigative reporter is more easily appeased than the mellowest engineer riding a Prozac high. Engineers don't accept intuition, aren't swayed by emotion, refuse to surrender to rhetoric, and can't let anything imperfect pass by without comment. Engineers never stop asking "Why?" until they get an answer they consider demonstrably, provably, irrefutably true.
As Craig Silverstein explained it to me, "It's not an engineering personality to keep quiet when you feel things are going wrong ... and being intimidated by people is not very productive."
"Google engineers were so strong-willed," agreed Matt Cutts, "that sometimes if we thought that Larry and Sergey were wrong, we just ignored them."
Many Google employees lacked nonacademic work experience. They had never tasted the lash of a manager's reproach or the sting of a colleague's rebuke, so their impulse to jump over fences and spread the wild oats of their wisdom remained unneutered.
"They would fight over everything," according to HR manager Heather Cairns, who often dealt with their issues related to job benefits. "'Why should I sign this? I'm not gonna sign it.' They challenged the most mundane things—minutiae—even though it was a benefit for them."
Project manager Deb Kelly, who got a steady stream of unsolicited feedback, wondered about those who offered advice, "Why are you weighing in? I'm not even sure I've met you yet."
This state of perpetual "Why?" occasionally annoyed even other engineers. Hardware designer Will Whitted found the ever-inquisitive commentariat a pain: "It may have been powerfully important in getting Google to work, but it pissed me off a lot. Because they were really, really bright and had been successful in one thing, they thought they automatically knew everything else. So they could tell me how to do thermal design and they could tell me what size screws to use. They honestly thought it would make things right or they thought the thing I was doing was incorrect and would make things bad."
Emerging unscathed from this idea melee required access to information about the company's inner workings. At first this information floated freely, permeating Google like radio waves. You just had to know what frequencies to tune in to. I took to poring over engineering's weekly snippets and diving deep into MOMA, our intranet. I paid attention at TGIF and eavesdropped on tablemates at lunch. No one consciously tried to limit data flow, but we lacked a formal clearinghouse for updates on company initiatives.
Larry's product-review meetings created a central information nexus. I could sit on the black couch, plug directly into Larry's head, and get root-level access to all that I needed to know. Nothing helped me do my job better than downloading directly from Google's wellspring of strategic direction. Cool draughts of clear vision washed away ambiguity about user interfaces, product features, and competitive positioning. I basked in my unobstructed view of the deliberations driving our company's creation, blissfully unaware that I would soon be banished from this information Eden and forced to forage for the info bits that I had come to rely upon to do my job.
Yes, I'm Saying No
With everyone expressing opinions about everything, I had to speak louder and more insistently lest my voice be lost in the din. Apparently other people felt the same way. We held robust dialogues.
When I needed to advance an argument, whether about the name for a new product or the wording of a promotional line on the homepage, I began by building alliances. I put Salar near the top of my list of draft picks, because he and others who had joined Google before the move to Mountain View* held sway in the decision-making process. Any of us could talk to the founders and have our opinions considered, but their voices were the first to be sought and the loudest to be heard. When one of this inner circle diametrically opposed my position, earning the counterbalancing support of another became essential.
Most flare-ups quickly damped to a simmer, the heat dissipated through flames on topic-specific email lists, the private clubs where Googlers could assemble to gnaw their own intestines. Particularly hot issues, however, might engulf the full Googlers list, which went to every inbox in the company.
"Engineers are argumentative," Urs acknowledged. "You don't want to stop it, but it shouldn't get in the way. At some point we made Googlers a moderated list,† just because people weren't thinking that their messages went to three hundred people."
"Googlers had passionate beliefs," remembered Matt Cutts, whose first pass at a porn filter was publicly picked apart on the engineering list. "So two entirely polite, consensus-driven people—you would have to send them on a walk around the block and have them work it out, because they both believed so strongly."
Matt believed the head-butting over issues outsiders wouldn't care about—if they even noticed them—wasn't about turf or politics. It was about determining what was best for users. The question, then, was who best represented the voice of those users within Google.
I thought I did.
I read users' email to us. I wrote the words that spoke to them from our site. I had spent decades in direct contact with customers of all kinds who didn't understand how to use any device with more than two knobs and a button. In UI team meetings and product reviews, it was easy for me to see which changes could cause confusion, because those changes confused me. I might not have been a digital everyman, but I was better cast in that role than my colleagues who could name a half dozen open-source operating systems and considered them all superior to Windows.
Marissa Mayer also strove to channel the concerns of users, referencing her mom in the Midwest as someone to keep in mind when introducing features or rearranging the interface. It was natural that Marissa's role as the interim UI engineer would lead her to safeguard user interests, and she did so with intensity. Her enthus
iasm and intelligence carried her opinions as a kind of rolling assault. If you weren't initially overwhelmed, she launched wave after wave of data, ideas, and arguments like landing craft at Normandy.
In a data-driven company, numbers are a big stick to wield, and Marissa cited stats that convinced her she was empirically correct. But while I had seen the power of quantitative analysis to persuade, I had never fully climbed aboard the data train. As a result, we didn't always agree.
If I lacked the numerological faith of my colleagues, I did share some of their characteristics: a focus that detected subatomic flaws, an inability to ignore inconvenient truths, and an obdurate unwillingness to cede my position until completely overrun. I stood up for what I believed, even when my only support was the gates of Hell pushing into my back.
I come from a stiff-necked people.
I often yielded on issues of design—or at least funneled my views through Karen, who had complete credibility as an impartial webmaster—but my disagreements with Marissa about wording and tone grew deeper after the MentalPlex April Fools' contretemps.
Many Googlers believed they spoke the tongue of the "average user" with native fluency, not realizing how thick a geek accent they brought to conversations about privacy or customer communications. What would users like? What would they find intrusive? Offensive? Why would anyone be upset by that? We all quoted the gospel of efficiency and swore to put our users first.
Whatever the nature of our opposing views, our culture urged quick resolution. Googlers wanted to get things done. Sanjay described engineers' approach to disputes this way: "We said, we can keep on discussing this for a long time and try to get agreement or we can just go ahead and do at least the part we know."
Complete buy-in wasn't a requirement. If there were holdouts, Urs would call a meeting and announce, "Okay, fine, we've argued for a week. There are no new insights being produced. Let's do the pros and cons and make a decision and move on. Because it's time to move on."
Larry was the only one who could play that role when it came to interdepartmental divergences, and given Cindy's directive not to delay product launches for marketing reasons, there was some risk in defending a branding perspective past a certain point. I had trouble seeing that point. When my lack of a self-preservation instinct became apparent, others sought my help in elevating issues to upper management. Would I forward concerns about our poor translation quality? Would I put the brakes on an off-strategy plan regarding Japan? Would I throw my body in front of a badly timed product launch?
Yes.
Yes.
Yes, I would.
The Sweet Taste of Porn
Larry and Sergey randomly fixated on details that caught their attention, such as the exact shade of yellow paint to be applied to the Google Search Appliance (our "search engine in a box" for corporate intranets) or the wording of their biographies on the website, but mostly they set up a management infrastructure, wrote a few rules, and let the system run. Each new layer of process would require compiling time and slow things down, so they promulgated a laissez-faire style that largely left employees to their own devices.
Total autonomy was a satisfactory state for most engineers. "Larry and Sergey went to meetings every week, sat in back, and listened to people talking about things," Matt Cutts later recalled. "They'd give us room to decide whatever we thought was the most important thing to work on." Which isn't to say that the founders didn't express their own strongly held beliefs. "If something didn't match their intuitions," Matt added, "they fought until they had good data or a good reason to believe or had seen a particular person be right a few times. Then they'd be willing to trust that person's judgment."
Matt earned his own chunk of autonomy by taking on one of the company's dirtiest jobs. He'd been at Google a month or two when project manager Deb Kelly stopped by his cubicle with a question.
"Hey, Matt. How do you feel about porn?"
"That depends," Matt replied. "Why are you asking?"
Deb needed someone to work on a filter to screen out "adult" content, which by definition meant being exposed to the seamiest parts of the web. Matt agreed to tackle the job, thinking it would take only a couple of weeks. Instead, it was three months before he had a prototype ready to test. To give it a thorough vetting would require conducting more searches than Matt had time to do himself, so in May 2000 he sent out a call to his fellow Googlers for help.
No one responded.
"It's weird," Matt said to his wife that evening. "You'd think people would take advantage of an officially sanctioned opportunity to look for porn."
His wife thought it might just be a matter of the proper motivation. "Why don't I make cookies," she suggested, "and people will get a little reward."
The next day Matt reiterated his need for help but augmented it with an offer of free "porn cookies" for everyone who participated. Search for porn, get a cookie. I was one of those who couldn't resist such a come-on. Over the next few hours I learned several new words and new meanings for some I thought I already knew.
Porn is a cutthroat business that often leads in the exploitation of new technology. Working on what would be dubbed SafeSearch, Matt became aware of a new problem for Google. Spam. Spammers attempted to game the system and thereby win a higher ranking in search results than they deserved. Matt came to think of spammers as "black hats" who placed invisible white text on a white background, stuffed their pages full of keywords, and employed a wide range of sophisticated and devious means to deceive Google's search bot.
"Once you start to see spam, the curse is, you'll see it everywhere," Matt told me. He was offended by spammers' unethical behavior and continued thinking about the problem even after he finished his filter and began working on Google's advertising system. Larry and Sergey thought spam was a non-issue, because they were confident Google's PageRank algorithm would sort the wheat from the chaff.
"It took quite a while for Google to wake up to fact that while PageRank was very spam resistant, it wasn't a hundred percent perfect," Matt noted. He took the initiative to sound the alarm. Although it was no longer his area of responsibility and not an area of concern to management, he asked to work on fighting spam. "One thing I learned at Google," he said to me, "is that you make your own cred. If you propose your own initiative, you're much more likely to do it than if you sit around and wait for someone to say, 'What do you want to do with your life?'"
Within days, Matt connected with the group of eight engineers focused on overall search quality, including Jeff, Sanjay, and Amit Singhal. They invited him to join them around the Ping-Pong table and delve into the deepest aspects of Google's core technology. Matt became privy to Google's secret sauce, the weighting factors that determined whether a website was near the top of the first results page or buried somewhere on page thirty.
Matt always struck me as a pillar of moral rectitude, a keeper of the faith in algorithmic integrity, and an adamant protector of Google's purity—our own avuncular Elliot Ness, bringing to account those who parasitically thrived by bootlegging traffic. As a University of Kentucky student, Matt had enrolled in a co-op program through the Department of Defense and ended up spending a few semesters at the National Security Agency. That internship was great fodder for the conspiracy theorists monitoring Google. Matt's "secret" government tie made their tin hats stand at full attention, because they assumed Matt still served as a conduit between big government and big search.* In truth, Matt was one of the least "spooky" guys I knew at Google.
Because he understood the innermost details of Google's ranking calculations, Matt was outraged when a webmaster bulletin board speculated in 2001 that Google was manipulating results to increase sales of our advertising. If a business didn't show up near the top of our results, they alleged, it would have to buy ads to have a presence on the first page users saw after a search. The rumor wasn't surprising, given the practices of most search engines at the time. The Federal Trade Commission had called out eight of G
oogle's competitors for blurring the lines between paid placement, advertising, and algorithmically produced results. Matt again took the initiative to address a problem others didn't immediately see. He went to PR manager David Krane.
"So ... while I'm compiling," he asked Krane, "would it be okay if I stopped by this forum and debunked misconceptions?" Krane reported to Cindy, and Cindy had read the Cluetrain Manifesto—a guide advocating that companies speak directly and plainly to the public instead of engaging in Velveeta-smooth, committee-processed, content-lite corporate blandishments. Cindy had made sure everyone else in marketing read it as well. Krane gave Matt carte blanche to speak freely on the company's behalf, without running his posts by PR.
"GoogleGuy" was born.
Matt's nom de plume became an authoritative voice in the webmaster community and a trusted source of information from inside Google. GoogleGuy corrected misinformation, killed rumors, and explained why Google did things that seemed off kilter to outsiders.
Initiative crossed with autonomy provided unforeseen benefits in unexpected areas. Unexpected challenges as well. Other engineers also sought to speak in unvarnished language on Google's behalf without formal endorsement by Cindy's minions—or even our awareness.
Ray Sidney comes to mind.
The Ray Way
Ray, Google employee number six, embodied the cult of individual authority. His "dude"-infused speech and ribald and unpredictable passions obscured an education earned at Caltech, Harvard, and MIT. Ray was our first line of defense against webmasters who pummeled Google with automated queries. Webmasters and SEOs* wanted to make sure their sites showed up near the top of Google results and so used monitoring software to conduct repeated automatic searches for keywords important to them. In periods of high volume, automated queries slowed down Google for everyone, which is why we considered them a violation of our terms of service.†
I'm Feeling Lucky: The Confessions of Google Employee Number 59 Page 18