In The Plex

by Steve Levy

  Buchheit figured, why not try it out? The two engineers sat down by Sanjeev’s computer and went through his email, trying to pick out a subject and then typing the word into the search engine to see which ads came up. They discovered that Google had plenty of ads in its inventory to produce ads relevant to at least some emails. Then he began figuring out how to automate the process. He downloaded a program from the web that used semantic analysis to distill blocks of text into a few keywords. Then he accessed Google’s AdWords system, replacing the keywords that advertisers requested with the keywords he’d extracted from the text analysis. When he was done, a series of sponsored ad links appeared to the right of the body of an email—presumably, linking to products relevant to the email’s content. “It was a very basic thing, and I implemented it in a few hours,” Buchheit says. (Eventually, Gmail would use the same semantic analysis system as AdSense, based on Georges Harik’s Phil project.)

  “It took people by surprise,” says Buchheit. “It was by far the most negative backlash of any feature release we’d done on Gmail.” The annoyed Marissa Mayer asked, “Why are you building this?” But Brin and Page thought the idea was cool and useful, an unbeatable combination. “We were really entranced by it,” says Page. “We really felt like, ‘Wow, something was mentioned in my email and I actually got an ad that was relevant!’ That was amazing. We thought that was a great thing.” As for the potential blowback, Brin says, “We didn’t give it a second thought. There were plenty of things to question, but I never batted an eyelash at that. It never occurred to me as a privacy thing.”

  Even Mayer came around. While testing the system, she was in an email thread where she was arranging a hiking date. Up popped an ad for hiking boots. This can actually be useful, she thought, and from that point she was on board.

  Caribou took forever to develop. Part of the problem was that Larry and Sergey were so invested in the project. They adopted it as their primary email system and would often drop by to give criticisms and suggestions. Buchheit would often take a working prototype to the weekly Google product strategy meeting, where product managers submit their products to a human wind tunnel of executive criticism. Products have been known to die at GPSs; there are stories of teams entering the conference room, exhausted and hopeful after long hours of getting a demo just right, and Page saying, “You’re wasting our time” and ordering the project dismantled. Larry and Sergey liked Caribou too much to kill it but dished out very tough love. At one point Page told the group, “I’d rather be doused with gasoline and set on fire than use your product.” But finally it was ready to be released in beta version. (Google often kept its products in beta much longer than other companies, signaling that users should be tolerant of faults and that an update was probably around the corner. In the case of Gmail, which became the public name for the project, the beta label was not removed until five years after Google released it, when it had tens of millions of users.)

  Brin and Page both thought Gmail was special, so they thought it appropriate to launch it on a day that was special for them: April 1. It was a definite stumble. When your main competitor allows only 2 megabytes of storage, people think you’re goofing when you boast a service with 1,000 megabytes and announce it on the day you usually unveil phony products. Even years later, Brin still relished the reverse spin—tricking people by not hoaxing. “I liked doing it on April Fool’s Day,” he says. “We learned some things, but that doesn’t mean that we don’t want to do audacious things on April 1.”

  The launch was problematic for a number of other reasons besides the question of whether the whole thing was a spoof. Even though Google was making a public announcement of its new product, the public couldn’t sign up for it. Google limited the number of users by declaring that it was by invitation only. At launch, it gave away 1,000 accounts to outsiders, allowing each new user to invite a few more people. Those left out were frustrated.

  But much more serious was what happened when people saw how Gmail operated. They were shocked when, alongside the text of emails, they found ads that seemed related to the content. It was as if Google were looking over their shoulder and snooping on their mail.

  A second, related complaint came from Google’s boast that with Gmail, you could keep your email forever. People were accustomed to having files of email on their own computers, in locations they could identify. And here was Google, the harbinger of a new age where everything—be it business confidential or searingly personal—would be stored on computers owned by Google that were physically God knew where.

  Those complaints hit Google’s engineers by surprise, because they wanted their mail kept forever and believed that in a connected world information was best kept in some futuristic version of safety-deposit boxes, maintained by professionals. They considered the privacy concerns illogical. They trusted machines, and their own intentions were pure—ergo, people should have trusted them.

  Inside the bubble of the Googleplex, that made sense, but the engineers failed to understand how, from the user’s point of view, Gmail was different. With Microsoft’s Hotmail and Yahoo’s Yahoo Mail, the low ceiling on storage meant that only a small fraction of mail would be kept by those companies. Google, on the other hand, would keep a comprehensive archive. And though it was true that Microsoft and Yahoo also automatically scanned mail in their systems for viruses and other things, users actually saw the evidence with Gmail. By serving ads related to content, Google seemed almost to be reveling in the fact that users’ privacy was at the mercy of the policy and trustworthiness of the company that owned the servers. And since those ads made profits, Google was making it clear that it would exploit the situation.

  But it wasn’t just Gmail that disturbed people. Suddenly, Google itself was suspect. Until 2004 Google had been seen as a feisty start-up performing an invaluable service. But it was sitting on a privacy tinderbox. One key issue was Google’s retention of user requests and responses when they visited its search engine. You can’t get more personal than that. A search history can reveal your health problems, your commercial interests, your hobbies, and your dreams. What would your health care insurer think about your search for “chest pains”? What would your investors think if you searched for “bankruptcy lawyers”? What would the cops think if you searched for “hydroponic equipment”? What would your spouse think if you searched for “afternoon sex encounters”? What would the government think if you searched for “tax resistance”? In 2006, the government, in a fishing expedition for information to help efforts to regulate pornography, would demand that Google and other search engines hand over logs of millions of searches. Google alone fought the subpoena. But when privacy advocates demanded that Google not retain any logs at all, the company balked. Those logs were the lifeblood of Google’s persistent drive to improve itself, the oxygen of its effort to become an unprecedented learning machine.

  In a sense, Google had been fortunate in postponing the inevitable privacy showdown until Gmail’s arrival. The excellence of Google search had been exposing personal information ever since Brin and Page first started spelunking in the caverns of the web. It wasn’t Google that put information on web pages and other online repositories, but it was Google that dislodged it all. Like it or not, Google had a role in the process. Though the issue hadn’t exploded into an outcry such as Gmail created, it had been simmering constantly. Larry Page recognized early on that “there’s going to be large changes in the world because of all this stuff,” and those Internet benefits might have a cost. “People will have to think when they publish something online, ‘This might be forever associated with me.’ Because Google exists.” But the problem wasn’t just what people published—Google was relentless in finding out everything about them, whether it was an address previously hidden in a database or a twenty-year-old article about a criminal charge that may or may not have resulted in a conviction.

  Denise Griffin, a Googler who joined in 2000 and worked in the tiny marketing department, was
the person charged with handling complaints. It was often heartbreaking to hear stories of how the things that Google dug up caused hurt feelings and sometimes caused actual harm to people. Google’s official stance, with some justification, was that it was simply delivering information that resided on the web. That explanation did little for those who felt exposed; without Google, all that information would have remained buried. “It’s very tough for people to conceptually understand that it’s not our website [exposing them], it’s the web,” she says. “We had a few different canned responses that we would send people, trying to explain, and when they would write back cursing, then we would try again with a slightly different version and then a slightly different version.”

  The worst situation was when someone was put into physical danger from information unearthed by Google—for example, people who had gone to great lengths to hide personal information from abusive ex-spouses and found their efforts undone in a 400-millisecond Google search. “I would feel terrible,” says Griffin, who would try to suggest remedies, such as contacting the webmaster where the toxic information lay. But unless there was a legal justification for removing the information—copyright infringements, child pornography, libelous information as determined by a court—Google said it couldn’t do anything. And philosophically Google was perfectly fine with not doing anything except in those cases. Brin and Page both believed that if Google’s algorithms determined what results were best—and long clicks indicated that the algorithms were satisfying the people who did the searching—who were they to mess with it? That was essentially the message they gave to Denise Griffin when she shared her concerns with them. “They were very frustrating conversations for me,” she says. “I lived this. It was really hard to get these emails.”

  It was a problem that Googlers themselves often saw firsthand. At one point, a search engineer named Jessica Ewing challenged the search team to do something about the fact that the first search result under her name was a mortifying picture of her as a thirteen-year-old Michigan all-state mathlete. “I will never have a date again!” she wailed. But there were less frivolous complaints. When you did a search for Google executive Susan Wojcicki, for instance, the second result was a posting from the Silicon Valley gossip blog Valleywag, inaccurately charging her with stealing the credit for developing AdSense. Wojcicki knew why Valleywag’s posting was ranked highly—“to link to a well-read article is not a crazy thing,” she says. But she didn’t like it. “Yes,” she would say when pressed on the issue, “it does bug me.”

  At least those Googlers understood that their status as employees should give them no privilege to censor the company’s indexes when other people could not. One day Denise Griffin got a call from Eric Schmidt’s assistant. “There’s this information about Eric in the indexes,” she told Griffin. “And we want it out.” In Griffin’s recollection, it dealt with donor information from a political campaign, exactly the type of public information that Google dedicated itself to making accessible. Griffin explained that it wasn’t Google policy to take things like that out of the index just because people didn’t want it there. After she hung up the phone, she freaked out. Doesn’t Eric know that we don’t do that?

  She called her boss, Sheryl Sandberg, and they had several conversations before they finally trudged up to Eric’s office and told him it wasn’t Google’s job—nor should it be—to filter his personal information. Griffin understood how he felt, because she came across upset people all the time. You could explain forever how making obscure but damaging information available in milliseconds was at the core of Google’s lofty mission. “Principles always make sense until it’s personal,” she says.

  Then in July 2005, a CNET reporter used Schmidt as an example of how much personal information Google search could expose. Though she used only information that anyone would see if they typed Schmidt’s name into his company’s search box, Schmidt was so furious that he blackballed the news organization for a year. “My personal view is that private information that is really private, you should be able to delete from history,” Schmidt once said. But that wasn’t Google’s policy.

  If Google’s own CEO had trouble dealing with privacy, how could ordinary people cope?

  Google’s Gmail conflagration needed dousing. Fortunately for Google, the company had recently beefed up its policy and legal team. Google’s original counsel, David Drummond, who came to the company from the big Silicon Valley law firm of Wilson Sonsini Goodrich and Rosati, preferred more of a business development role to a purely legal role, and he hired a group of people with experience in protecting civil liberties to help shape and defend Google’s policies.

  A big part of the task fell to Nicole Wong, an attorney who had been hired only four months before the Gmail release. She had a joint passion for news and law—she earned a master’s in journalism at the University of California at Berkeley while getting her law degree there—but eventually decided on the law. She sent résumés only to law firms with First Amendment practices, winding up at a San Francisco firm with media clients. In 1997, she began to move into Internet-related issues and moved to a Seattle firm that did some privacy work for Google. At a breakfast meeting with Drummond one day, she was pitching her firm for a bigger role at Google when Drummond asked if she’d like to do those things as an employee. The carrot was the ability to write her own job description, which turned out to be a list of legal problems that would trouble Google for the rest of the decade: intellectual property challenges, defamation, invasion of privacy, and content regulation.

  Starting at Google on the same day as Wong was another lawyer named Andrew McLaughlin. With a background in Internet administration—he’d worked for the Internet governance organization ICANN, requiring him to deal with a passionate constituency of geeks and freaks—he became Google’s first policy director. He didn’t know much about Gmail until the day before the product release, when the whole company was summoned to Charlie’s Café for a demo. Everyone left with a Gmail account and a T-shirt. McLaughlin left with a headache. Oh, crap, he thought, I better get in on this.

  The reaction was all that he feared, best summarized in a CNET headline: “Why Gmail Gives Me the Creeps.” Simon Davies, the head of a group called Privacy International, put out a press release saying, “Google looks at privacy the way a worm looks at a fishhook.” Google had gone from cuddly Internet icon to Big Brother in one day. From that point on, instead of the introduction McLaughlin had hoped for when meeting legislators and industry groups—“Hi, I’m Andrew from Google, let’s talk about policy”—he had to use a different opening line: “Hi, here I am, and I’ve got to explain this thing that seems creepy and weird and convince you it’s not so bad.”

  When the outburst hit, Page and Brin called for a war room. Buchheit’s troops were joined by engineering teams, PR people, and Google lawyers. Brin got on the phone to a sympathetic writer at Salon.com. “We will use good judgment,” he promised, after laying out why people should not be outraged.

  The most serious challenge came from a California state senator from Fremont, barely out of Wi-Fi range from the Google campus. Liz Figueroa, McLaughlin would later recall, was exploring a run for lieutenant governor and was on the hunt for issues. One of her top staffers had become a father a few months earlier, and he and his wife had begun receiving free samples of baby lotion and other products. The staffer had been horrified that corporations had used personal information to solicit him. “It made a real jihadist out of this man,” says McLaughlin. The staffer clearly saw Google’s new product—it was reading citizens’ email!—as a threat to society.

  Figueroa appears to have had a minimal grasp of technology and apparently didn’t realize or didn’t care that many of her constituents worked at Google. She introduced a bill that would ban advertising targeted to email. “Telling people that their most intimate and private email thoughts to doctors, friends, lovers and family members are just another direct-marketing commodity isn’t the way to p
romote e-commerce,” she said in a statement.

  But McLaughlin had a card to play. “I mobilized the Big Al,” he says.

  That meant Al Gore, the former vice president. In the years since losing, or maybe not losing, the 2000 election, Gore had taken comfort in a warm reception in Silicon Valley. “I was trying to figure out what the hell I was going to do with my life,” he says of this period. His geeky side led him to Page and Brin, and Google asked him to join its board of directors. But still undecided about another campaign, he was avoiding such offers. (Later, when he’d made up his mind not to make another run at the White House, he joined Apple’s board.) But at Google he agreed to become a “virtual board member,” with the formal appellation of senior adviser, consulting with the five or six top leaders at Google and occasionally helping pull a lever or two with a government contact. McLaughlin, who didn’t have much clout with Larry and Sergey, spoke to Gore regularly and would sometimes implore the former VP to talk to the founders when it looked as though they were about to make a wrong call on a policy issue. In this case, McLaughlin asked Gore to speak to the Democrat state legislator who was giving Google a privacy hotfoot.

  Figueroa agreed to meet with Gore at the Ritz-Carlton in San Francisco, where the former vice president was staying. Gore was ready for her. He launched into a defense of Gmail that was nearly as elaborate as the climate change slide show that would later help him share the Nobel Peace Prize. “He was incredible,” says McLaughlin. “He stood up and was drawing charts and did this long analogy to the throw weight of the ICBM, the Minuteman missile.”

  “It was a full and candid discussion,” Gore would later recall, while claiming not to recall the ICBM analogy. (He did cop to the use of “a whiteboard, large-scale Post-it notes, and diagrams.”) “We talked through the fairly complicated nature of the advertising model that had the robotic analysis, without giving any human being any access to email.” Eventually, Figueroa modified her bill to permit the kind of automated scanning Google performed in Gmail.