Missed Management
I saw Eric often in his first weeks. He seemed to spend much of his time roaming the halls with a bemused look on his face, as if he couldn't believe he'd actually joined this company populated with big rubber balls and lava lamps and scruffy animals sleeping on couches—sometimes with the pets they had brought to work lying next to them.
Usually when I saw Eric he had company. One day it was Governor Howard Dean. More than once it was Al Gore. Gore apparently had plenty of free time on his hands. I ran into him everywhere.
"Good afternoon, Mr. Vice President," I said to the tall man standing at the urinal next to me as I took a break between meetings. That experience pretty much dissipated any residual sense of awe.
When Eric and Al stood outside my office chatting about Gore's interest in creating a new independent TV network, I ever so gently nudged the door closed with my foot. I had work to do.
I wasn't sure about Eric. No one seemed to know what he was doing, since Larry and Sergey were still making all the decisions. Other than the purchase-order edict, we didn't hear much from him for a while, perhaps because he didn't officially add the title CEO until August 2001.
Eric did add transparency to the decision-making process, forcing discussions in public meetings so everyone could see the sausage being made, even if we didn't always like the ingredients stuffed into it.
"In a culture which is consensus driven," Eric explained to reporter Fred Vogelstein in November 2005, "the trick is to have everybody participating in the decision and make sure everybody has been heard."* That wasn't true at Google before Eric came onboard. As one drifting into a more distant orbit around the stars of our universe, I appreciated Eric's efforts to shed light on a process that all too often took everything in and let nothing out. That support for transparency led me to view him as an ally, a friend of marketing, and a voice of reason.
A month before Eric could implement his ideas as CEO, Google's top-down decision-making reached its peak—and its nadir. Larry decided that July to reorganize the engineering group. It didn't go well. Our founders cut bureaucracy the way they cut costs—with a cleaver instead of a paring knife.
Larry trusted his newly minted product managers (PMs), Salar and Susan, and the two new hires, John "J.P." Piscitello and Pearl Renaker, who joined them in April 2001. They all reported directly to him. He was less happy about the half-dozen project managers who had been working with engineering for months. Project managers created timelines, allocated resources, and prodded engineers to ensure products shipped on schedule. They conducted performance reviews and kept people like me from bugging the technical staff with requests that might slow things down. They also acted as a buffer between the engineers and the executives, or as systems engineer Ben Smith described it, "shielded their employees from random shit that came from above." In a company like Google, standing between the founders and the engineers did not earn you a sash and a spot on the homecoming float.
I liked the project managers. More important, I needed them. They were our contacts for any task requiring technical resources, such as evaluating vendors or tracking the performance of our banner ads. The project managers assigned engineers to work with us—engineers who were under great pressure and would not normally consider a marketing task important enough to add to their to-do lists.
Wayne Rosing joined Google in late 2000, and in January 2001 took over as VP of engineering, replacing Urs Hölzle, who, as the first Google Fellow, went off to solve large-scale operations issues caused by our rapid growth (things like energy consumption and efficiency across hosting centers). Wayne had an avuncular manner and a Charlie Brownish appearance that he subverted by occasionally sporting a diamond-stud earring or dying his hair bright red. He brought with him decades of experience in running engineering groups, including Sun Microsystems Laboratories, which he had founded. Wayne discovered that Google engineers largely controlled their own destinies, sometimes acting on—and sometimes ignoring—priorities that flowed from Larry and Sergey in an ever-shifting spectrum of urgency.
The company grapevine soon grew heavy with rumors that big organizational changes were coming, fermenting the staff, who whispered nervously about layoffs and what that would signify at a company not yet out of its infancy. Still, when Wayne called an all-engineering meeting in July 2001 and announced a reorganization, most of the engineers were caught by surprise. So were the project managers, who learned in public that their jobs no longer existed.
When the announcement was made, there was audible grumbling among the assembled engineers. They generally respected the project managers and felt they had a real role to play. And they objected to the idea of anyone's dismissal by public firing squad.
To stave off open revolt, Larry stood at the front of the room and laid out all the things he wasn't happy about in the engineering management system, starting with the idea that non-engineers were supervising those who knew more about the technology than they did. The remarks stung the project managers, some of whom felt it was a personal repudiation, especially since Larry had not raised the issues with them individually in advance of the meeting.
"It sucked," one of the project managers told me later. "I felt humiliated by it. Larry said in front of the company that we didn't need managers, and he talked about what he didn't like about us. He said things that hurt a lot of people."
The grumbling got much louder. "I yelled at Larry," engineer Ron Dolin admitted, "because he said that the managers they were planning to lay off weren't doing a good job. And I said that this was no place to give a performance evaluation. Laying these people off was completely ridiculous, and the nature of the announcement was totally unprofessional."
"I did my best to advise that there is true value in management," Stacy Sullivan recalls, "and you can set a tone by how you manage this. And Eric said, 'Let them try this.' Wayne said, 'Let them try it, it's their company.' It bothered me. We tried to give input—me, Omid, Urs. But in the end, hopefully it was a lesson learned for Larry and Sergey."
The solution that Larry wanted was to have all the engineers report directly to Wayne. While it was positioned as a way of streamlining the engineering structure, most of those I talked with thought it was really about Larry's priorities not being addressed.
"Larry and Sergey had certain things they wanted worked on," Gmail creator Paul Bucheit explained, "and there were these standing groups that were making up their own things and not doing whatever it was Larry and Sergey wanted." For example, Larry wanted to scan books. Many, many books. Every book in the Library of Congress. But no one seemed interested in undertaking such a wildly ambitious project. With the engineers operating as autonomous units under the protection of their project managers, Larry found himself increasingly frustrated.
Howard Gobioff was convinced that "this was about people getting between Larry and Sergey and the engineers. At a time when the organization was small enough that the founders still wanted to be very hands on. It was very badly handled. Most of the engineers were pissed because we liked our managers. They were non-technical, so they lacked delusions that they knew better than we did."
Urs, characteristically, blamed himself. "What caused it was my inexperience at managing," he told me, "and Larry being very good at recognizing the long-term conflict that created." Urs had believed his engineers would cover Google's coding needs, so he probed potential project managers for organizational ability and tested their people skills instead of their technical knowledge.
"So ... what I underestimated," he went on, "is that managers always make judgment calls. They have to in order to function. If you're in a highly technical area, you can't make good judgment calls if you're not highly technical yourself. We changed at that point our strategy for hiring managers—away from coordination to saying that what matters most is technical leadership."
Larry recognized the problem sooner than Urs did, but neither had the experience to make the transition graceful and pai
nless. Instead, Larry just did what came naturally. The system didn't work, so he rebooted it.
"I can't think of anything that people at Google were ever so upset about—at least in engineering," Paul Bucheit recalled years later. "people had some sense of ownership of the company, that it was this big happy family. And all of a sudden, some of your friends were kicked off the island. You're like, 'This isn't what I thought it was. I thought we were all in it together, and we just decided to get rid of these people.'"
At most companies, the notion of an engineering head having hundreds of direct reports would be ludicrous, but because he believed Google engineers were self-directed, Wayne just did away with the management layer between him and them. He divided the engineers into teams of three, with each team having a technical lead who was an engineer, not someone hired to manage.
The catch was that each team would also have a manager assigned from Larry's new product organization. It was a not-so-subtle introduction of a true product-management system. The project managers who had covered the engineers' backs had been replaced by Larry's trusted lieutenants, who would be looking over their shoulders.
"You could do a lot of stuff with tech leads," search quality expert Ben Gomes explained to me, "because of the people we hired. Anywhere else, having three or four hundred people report to one person would have been insane. Yet it worked reasonably well—for a while. And then at some point it didn't work."
Ultimately, the project managers were spared. Urs absorbed most of them into his operations area. But the angst unleashed by the reorg did not fade quickly.
When the dust settled, all hundred and thirty engineers reported directly to Wayne. The bureaucracy was dead. There was no hierarchy. There were no in-depth performance reviews. Engineers were on their own, independent entities, connected only to the other members of their teams and tenuously tethered by PMs to the central organization. Their direct interaction with Larry happened mainly at product review. Wayne took to holding weekly meetings and to walking through the cube farms on a regular basis to ensure that he had face time with individual engineers and that they were able to approach him with issues that concerned them.
It was an engineer's dream come true or a bit of a nightmare, depending on whom you asked. No clueless pointy-haired boss could get in the way and screw things up, but there were no clear signals from above about what was important and what was urgent and what was both. Groups struggled for resources and fought redundancy. Some engineers wanted more feedback on what they were doing and how well they were doing it, and others wondered about opportunities for advancement.
The true significance of the reorg would not be immediately apparent, because shortly after we began rebuilding our world, the rest of the world fell apart.
Chapter 16
Is New York Alive?
AS I DROVE to work on September 11, 2001, my mind was on Ask Jeeves. Late the night before, Larry had sent around a Wall Street Journal article announcing that our competitor was buying Teoma, a promising new search engine. That worried me. The Jeeves brand was strong, though their search technology couldn't compare to Google's. If they actually improved it, they might become a formidable player in the industry.
On the car radio, I heard something about a plane crash in New York City. I envisioned a Piper Cub that had been sightseeing and gotten too close to a skyscraper. And then they were talking about another plane. Another plane had flown into the World Trade Center. Jet planes, filled with people. The World Trade Center was burning. People were jumping out of windows. Other planes were missing. No one knew what was going on.
By the time I got to work, the TV was on in the blue conference room. A couple of engineers sat transfixed, bowls of soggy cereal untouched on the table in front of them. I sat down at the table to watch and didn't move for thirty minutes. "Oh my God," I thought. "Oh my God." I didn't think much beyond that. It didn't occur to me that there might be something I could do about what I was seeing. The disaster was on the other end of the country, three thousand miles away. I never once considered that I worked for a powerful global information service—that Google could somehow offer assistance.
Sergey walked in. He was frowning and clearly agitated. But his mind was clear. He saw problems and it never occurred to him that we could not help—that we would not help. He had been having trouble accessing online news organizations. People desperate for information had besieged them and choked their servers. He directed us all to begin downloading the HTML for news reports from whatever sources we could access. He wanted the text and the images too. He had already spoken to our webmaster Karen and to Craig, one of the few engineers who could manually push changes out to our website. We would harvest whatever pages we could and host them on www.Google.com, which was better able to handle high volumes of traffic than the New York Times or CNN.com.
No one asked whether it was within our legal rights to appropriate others' content. We didn't debate whether linking to cached news reports fit our brand, our mission, or our role as a search engine. No one argued that the links would disrupt the aesthetics of our homepage. People urgently needed information and couldn't get to it. We could help them. Clearly it was our responsibility to do so.
I realized then how much Sergey saw Google as an extension of himself. It wasn't an anonymous corporation bound by industry traditions. He had created it with Larry, and the only rules that applied were the ones they agreed upon. As William Randolph Hearst and Joseph Pulitzer had imposed their personalities upon their newspapers, Larry and Sergey had imprinted Google with more than just lines of code. The difference, though, was that Google's founders used the power of their "press" to present not just their own viewpoints, but all viewpoints.
I went back to my desk and checked my inbox. "Is New York alive?" read the subject line of a note from Chad.
"Oh, Christ," I thought. That's right. Google had an office in midtown Manhattan. Eric Schmidt was supposed to be visiting there that morning. The answer came back from New York that everyone was okay. They had evacuated their office near the Empire State Building when the first plane hit. They were shaken and they were concerned about friends and family.
My brain finally unfroze, and I thought about what Sergey was trying to do. I realized it would be better to host real news content than to put up a random collection of bits downloaded from across the web. I contacted Martin Nisenholtz, head of the New York Times's online service, and asked if he wanted Google to host copies of the pages they were posting. Martin was grateful for the offer but, after checking with his webmaster, declined. They thought their servers would be able to bear the load.
Meanwhile, Karen had assembled articles from the Washington Post and CNN and put them up on a page at Google.com/currentevents. We needed a pointer from our homepage, so I jotted down a paragraph and gave it to her. It read, "If you are looking for news, you will find the most current information on TV or radio. Many online news services are not available, because of extremely high demand. Below are links to news sites, including cached copies as they appeared earlier today."
I didn't think deeply about the implications of what I had written, which would be picked apart and sniffed at in the months to come as an admission that "new media" had not yet supplanted the old in a time of national crisis. That wasn't my intent, but I knew that our web index was not updated in real time. Searches on Google would fail to bring up any recent news, so the best we could offer users was links to news sites that appeared to be functional plus copies of static reports. It seemed obvious to me that live televised images from New York would be the most informative window on what was happening right at that moment.
The mainstream media, however, were not so quick to write off the power of Google. Within half an hour after the message went up, ABC News had asked for a link to their site. Then came MSNBC. We added them both, and in so doing exceeded the character limit for our HTML table, breaking Google's homepage. Karen and Marissa scrambled to fix it as the mood of
the day swung between depression over the events, obsession with garnering every scrap of information, and stress over trying to devise more things we could do quickly to help.
We fed relevant keywords into the advertising system so that searches on topics like "World Trade Center" displayed messages linking to our news page. We prepared a link to the Red Cross to encourage blood donations, then discovered that their site had been overwhelmed and couldn't be accessed. Our engineers ran a special crawl using the new incremental indexing tool we had been building for the Yahoo deal, so searches would bring back up-to-date results from news sites.
We were soon flooded with email from well-meaning users who wanted Google to help them help others. Most of the mail was about the ad hoc news directory we had created. Requests to be added to the list of links increased with each update we pushed out. A webmaster had a site where people could post messages letting others know they were safe. A Wiccan wanted a pointer to her "online healing book." An old friend at Salon.com wanted a link to their coverage. A British user suggested we add sources outside the United States. I explained to all of them that our mission was not "to replace news services online, but to help people get info they can't get otherwise." Still, the cascade of incoming links became a cataract, roaring in concert with the rush of the news from New York and the Pentagon and Pennsylvania.
I agreed about the value of adding a global perspective, so I asked Googlers what sites they used abroad. They sent back sources in German, Dutch, Swedish, Norwegian, Danish, Romanian, Polish, Spanish, Basque, Ukrainian, Japanese, and Russian. I dutifully checked them all. Unfortunately, I couldn't read any of them, so I had no way to evaluate whether they were espousing extremism in their bold headlines or had bias buried in the tiny type below. I tried to get confirmation from at least two Googlers before passing the sites along to Karen to post. The engineering team sent me a list of the top news sites they were seeing in the search logs—an indication of what sources people around the world were trying to find. If I couldn't get validation from staff, I took search popularity as a vote for legitimacy.
I'm Feeling Lucky: The Confessions of Google Employee Number 59 Page 27