I felt the truth of that deep in my soul. Those two sentences sparked an epiphany in me, that we could be the company that never tried to sneak an unpleasant truth past its customers. The company that always went overboard to be completely transparent about its actions. The company that did no evil. I would advocate as strongly as I could that we engage the issue of privacy and educate our users about exactly what information we collected. Surely that aligned with our core values.
When the Google toolbar launched in November 2000, the installer included the language as I had written it, bold red font and all. The page users saw after installing the Toolbar also called attention to the information collected and offered links to Google's privacy policy and instructions on how to disable the reporting function. People noticed.
Instead of a hue and cry there was a yawning silence. When users emailed us with concerns, we politely quoted the language on the installation page and reiterated that we had fully described all the options available to them. When the issue bubbled up on message boards, users who had already downloaded the Toolbar pointed to the bright red message and asked what more a person could want regarding notification. Those who downloaded the Toolbar with advanced features made a fully informed choice. Carnegie Mellon professor Randy Pausch confirmed it with a user study that led him to conclude, "The red Yada Yada message definitely works: it catches users' attention, and in a positive way."
For years, people talked about Google as a brand that could push hard against edges Microsoft could never approach without setting off alarms. The Yada Yada message was a big part of that.
News organizations from CNET to MIT Technology Review to USA Today and the Washington Post referred to Google's disclosure language in articles about the company and users' privacy, always giving Google the benefit of the doubt because it so clearly went out of its way to inform users about its intentions.
I believed that we had discovered the golden rule of user communication, and the cynical marketer in me rejoiced. We could defuse any controversial issue by rolling out our secret weapon: a brilliant and devious strategy that happened to be built on absolute honesty. It really was the best policy.
I may have taken that policy too far. With each new product we introduced, I refused to support grudging admissions about things users might have some interest in knowing. I demanded in-your-face, kimono-wide-open, wart-revealing, naked-before-the-eyes-of-God, full- frontal fact-flashing. I had been converted like Paul on the road to Damascus by the lightning bolt our bold red letters had deflected from our young brand. The more you informed people, the more they trusted you not to abuse them. Just as with MentalPlex, the lesson seemed as clear and sharp as shattered crystal. Yet, once again, others considered the same data and saw things in a different light. The issue of user privacy wouldn't arise again for months, but it would never cease to come back, no matter how many times we ignored it.
De Parvis Grandis
"This is version X.X of the Google toolbar," I wrote. "Earlier versions, if they exist, retain only sentimental value. Such is the transience of material things." Eric, the Toolbar engineer, was pleased. He had expected something more traditional for the text users and checked to see if they were up to date with the latest software. He soon added his own personal signature to the information box: "De parvis grandis a cervus erit" (Small things make a big pile).
It didn't always go so smoothly when I wrote copy at engineering's request. For example, we went back and forth on the text to display when users rolled their cursors over the "I'm Feeling Lucky" button on the homepage.
That button was an anomaly. If users clicked on it after typing a term in the search box, instead of seeing a full page of search results they would go directly to the one that would have been first on the list. Larry and Sergey were so confident in their technology that they thought the first result would provide adequate information for the majority of searches. Unfortunately, most of our users had no clue what the "I'm Feeling Lucky" button did. They let us know, however, that they liked seeing the phrase on the homepage and liked the self-assured attitude it implied. So we kept it—a small triumph for sentiment and brand building over cold, unemotional efficiency.*
In December 2000, I proposed "Go to the highest-ranked result" as the text explaining "I'm Feeling Lucky." Bay liked the way that emphasized our quality. Salar liked that it educated users about our unique method of ordering pages. Marissa thought it stank. Literally.
"Rank is a harsh-sounding word that has an alternate definition of 'bad smelling,'" she pointed out. She preferred "Go to the first result," because she wasn't sure users thought about search results as being "ranked." They more likely thought of them as "scored" or "ordered." You say tomato, I say tomahto. We ended up not implementing any rollover text at all.
I received similar pushback when I sent text to Schwim in ops for the automatic reply our system emailed to anyone who wrote us. He sent it back. With edits.
Schwim viewed himself as a sort of catcher in the rye, the last pair of hands to touch our outbound communications, and he took seriously his self-assigned duty to keep Google's brand from tumbling into the bottomless pit of mediocrity. He rewrote copy if it didn't meet the standards he felt Google should maintain. When a glaring typo mysteriously appeared overnight in an outbound email I had written, I confronted him. It hadn't been his fault. Another engineer had requested a change and introduced the error in the text. I had been copied in their online discussion, but hadn't replied, perhaps because the conversation took place at one-thirty in the morning.
Years later, Schwim explained. "We had a major product launch," he reminded me. "We weren't going home. We weren't going to bed. We wanted to give feedback and there wasn't anyone around, because surprisingly, you were sleeping." He meant it facetiously, but it was also true. The engineers weren't terribly sympathetic to the fact that I had a family at home and a life outside the office. Or that my day started at six a.m. while theirs ended then. Either you were available when they needed you or you weren't—in which case they would use their best judgment and move ahead. Nothing could stand in the way of the company's progress.
Maybe that's why Cindy kept my promotion to marketing director a secret, telling only those within our group. I'm sure it was a tough sell to Larry and Sergey to add a layer of management to the organization, even though December 2000 marked our first profitable month. What message would a director-level position in marketing send to the rest of the company? Would it indicate that marketing was gaining importance? I was doing my job and what was left of Shari's, but I wasn't writing code. I chose to view the promotion as a positive step. I had created a brand tone and put our external promotions in order. I was now the voice of Google. Cindy, at least, felt that was worthy of some recognition. As 2001 began, I looked forward to a new year of interacting with our users. My new challenge would be finding a way to answer the many, many questions they were starting to ask us.
Chapter 14
Googlebombs and Mail Fail
THE YAHOO DEAL turbo-charged Google's growth. Claus in our logs group carefully and scientifically plotted our traffic in crayon on a three-foot-tall roll of paper taped to one of the hallway walls. Significant milestones were precisely noted, with different colors signifying the components of our audience (searches on Google.com vs. searches on our partner sites like Netscape and Yahoo). Light- green foothills rolled in front of jagged dark-green peaks, beyond which could be spied a towering orange mountain range that stood in front of purple Alps majestically overshadowing all the others. The summits represented Mondays, and the valleys on either side reflected the weekend dips in our traffic.
Googlers annotated the chart with their own hire dates and other events of significance (such as the founding of the Google wine club). Celebrity-guest signatures appeared with regularity. Al Gore was there. Jimmy Carter. Chris Martin from Coldplay and his girlfriend Gwyneth Paltrow. Claus quickly decreased the scale of the graph so that three hundred thousan
d searches no longer appeared as a hill reaching halfway up the page. That mark now represented three million searches. Within months, he recalibrated so that the same height equaled thirty million searches. And still the mountains grew until they scraped the top of the chart.
Schwim sent out emails announcing major increments in the number of searches per day on Google.com, which started at around four million when I was hired in late 1999. Less than a month later that number, not including partners, was more than five million. Two months later we crossed eight million, and by the middle of September 2000 we were conducting a million searches an hour. Ninety days after that, it was a thousand searches per second. Claus added taller pieces of paper to the side of the graph as the volume of our traffic marched inexorably upward and to the right.
Operations had shifted into high gear on machine construction for the Yahoo deal, and now the production pedal was nailed to the floor. Fleets of computers provided enough capacity that a whole data center could be taken offline without bringing down Google or its partners. When a new index was built, the techs copied all the files to one data center and then moved on to update the next one. That caused some weirdness in our results.
"For about a week each month," Matt Cutts recalls, "depending on randomness and chance, you'd either go to the old data center or a new data center." If your search was routed to a data center that hadn't been updated and then you did the same search again at one that had, your results would appear to jump around, as if they were dancing. The "Google Dance" became an event of great concern to those who cared about their ranking in search results. On the WebmasterWorld website, commenters named each index switchover as if it were a hurricane, and the outcome could be equally devastating for any business whose website dropped off the first page of results. For Google, the switchover was not just about adding more data to the index. It was a time to tweak the ranking algorithms as well.
Google's focus moved back and forth from search quality to infrastructure. The initial innovation of PageRank at Stanford improved search relevance. The systems work in early 2000 led to the billion-URL index and faster response rates. In 2001, the pendulum swung back to search quality.
"Larry used to say, 'Search is too cheap,'" Urs told me. "The cost per query is really too low. Given the revenue we have, we should be able to do much more." Spending money to improve search quality would create a perceptible gap between Google's results and those of our competitors, enabling us to build a brand based on quality. Other search companies would have to invest at an equal or greater rate just to catch up. We would launch an arms race and spend our opponents into bankruptcy.
"Raising the cost of search will in fact increase the profit," Urs believed. "Your top line is going to go up much faster than your bottom line, and your margin is really very large."
Larry and Sergey wanted a focus on efficiency, but they also wanted to find out how much better Google could get if we threw twice as many machines at the problem of search quality. So they tried an experiment.
"In 2001," Sanjay recalls, "I was sitting in Larry's office and he just said, 'Here are three thousand machines. You guys figure out something to do with that.' It wasn't like we just happened to have all these spare machines lying around. It was driven by the need to figure out how to use lots of computation to do interesting things." That was just the beginning.
"There was an intent to send a signal," Urs confirmed. "Let's assume machines are plentiful—what could we do?" The three thousand machines were followed by ten thousand more, all allocated to improving search quality. Jeff, Sanjay, and Amit Singhal* focused all that computational power into an explosion of innovation that greatly improved Google's ability to return relevant results. One key to better results, according to Jeff, was "being much more careful about how we handled anchor text."
The more pages in an index, the more information there was about who linked to whom. And just as important to Google, the more information there was about what those links said. The wording of an actual hyperlink is the anchor text. Associating anchor text with the content of the page it pointed to turned out to be enormously helpful. For example, some links to the University of California at Berkeley contained its Spanish name or synonyms like "Cal" or "Bears." Sometimes, however, relying too heavily on anchor text caused problems.
"Some anchor text was good and some was not so good," Jeff explained. "The query 'cold lemon soufflé' used to bring up MapQuest's homepage." One website used that exact phrase in a link to MapQuest, and Google gave the association too much weight. Another example was "more evil than Satan." In 1999 that search brought up Microsoft's homepage.* Sergey instructed Cindy to tell reporters it was the result of an "anomaly caused by quantum fluctuations in web space," a nonsense phrase that was repeated verbatim in stories purporting to explain the glitch.
Sometimes deceptive link language was intentional. Webmasters realized they could affect the ranking of results for certain search terms by intentionally pointing to pages with very specific anchor text, a trick that came to be known as "Googlebombing." The Googlebomb that caused me the biggest headache was "dumb motherfucker." Around the time of the 2000 presidential election, conducting a search for what we euphemistically called "DMF" brought up an online store selling George Bush campaign merchandise. As the person responsible for customer service, I held up the umbrella as crap rained down on us from outraged supporters of the president. I got in touch with the Bush website people and explained that it was not an intentional slight by Google. Then I drafted a message saying the same thing for our customer service representatives to use as a reply to users. The best reply, though, came from someone already familiar with the issue.
A couple of months after the Supreme Court ended the 2000 election, Eric Schmidt invited his friend Al Gore to speak at Google about the campaign, the Internet, the environment, and whatever else Gore wanted to talk about.
Gore zipped into TGIF on one of our many electric scooters. "I just rode in from Washington," he said to approving laughs as he dismounted and took the microphone. "You may know me as the man who used to be 'the next president of the United States,'" he joked, garnering thunderous applause. When question time came, a Googler asked Gore if he was familiar with the Bush Googlebomb. He was.
"So what's your opinion about it?" came the follow-up question.
Gore paused and looked around a bit, as if checking for camera crews or reporters. There weren't any, though the room was packed with Googlers sitting in folding chairs and standing on cubicle desks to get a better look.
"Well," he said at last, with a completely neutral expression, "I do believe you might have discovered artificial intelligence."
Users would orchestrate many more Googlebombs in the years to come, from "French military victories," which led to a fake Google error page that read, "Your search—'French military victories'—did not match any documents. Did you mean French military defeats?" to "out of touch management," which, immediately before Google's IPO, directed users to the page on Google.com featuring profiles of our own executive team. That alone should prove to cynics that Google does not manipulate its search results or inject bias on a case-by-case basis.
Wait a Minute Mr. Postman
Each time a Googlebomb detonated it blew my schedule to hell. I'd have to drop everything to smother the flame wars that might damage our brand. User support cast ominous shadows across my world, but the poor grunt who faced the unending deluge of incoming missives was our sole customer service rep, Max Erdstein.
Fresh out of Stanford with a degree in history, Max had offered to do whatever needed to be done at Google, which initially meant working on writing projects like the Google Friends newsletter and responding to user email. After Cindy restructured the marketing group, Max moved into my world, dragging behind him responsibility for Google's rickety user-support department, which at the time consisted of Max, a laptop, and an off-the-shelf copy of Microsoft Outlook.
Max never envisioned
customer service becoming an omnivorous blob consuming all his time, but soon he found himself responding robotically to more than a thousand emails a day from users around the world. Crushed under the load, he could do little more than succinctly reply, "Thanks! Keep on Googlin!" Non-English emails presented the biggest problem. We had no idea if people wanted to praise us or harangue us. We tried using off-the-web translation software, but it left us more confused than when we began.
Meanwhile, there were rumblings from sales VP Omid that supporting advertisers and search-services customers should be a higher priority. Could Max help with that, too? After all, unlike users, these people were actually paying us. Max was emptying an ocean with a teaspoon. As the backlog of unanswered emails began to swell, Sergey offered a useful perspective. "Why do we need to answer user email anyway?" he wanted to know.
To Sergey's thinking, responding to user questions was inefficient. If they wrote us about problems with Google, that was useful information to have. We should note the problems and fix them. That would make the users happier than if we wasted time explaining to them that we were working on the bugs. If users sent us compliments, we didn't need to write back because they already liked us. So really, wouldn't it be better not to respond at all? Or at best, maybe write some code to generate random replies that would be fine in most cases?
Given their lack of concern about unanswered email, the founders were not sympathetic to Max's distress over the monotonous and unfulfilling nature of his work. Besides, when Google started, Larry and Sergey had answered all the email themselves, and written the code, and designed the logo, and handled press inquiries. We had more resources now, so how could anyone complain, given that we all had such small shards of responsibility in the wake of the big bang that was Google's conversion from a two-person project to a full-fledged corporation?
I'm Feeling Lucky: The Confessions of Google Employee Number 59 Page 24
