by Steve Levy
As explained by Mike Jones, who had come to Google as an executive of Keyhole, the satellite mapping company Google had purchased in 2003, Street View, emerged as a consequence of the omnivorous hunger for geodata. “From the day we came to Google, our constant petition was to get more money to buy more data, because we wanted to get the experience of seeing your home for everybody on the planet,” he says. “They’re going to want to fly in the middle of the Congo and see their house or their hut or something. And we needed to get pictures of that. We’d go to a GPS and say our goal is to gather this much imagery, and maybe take the crazy step of putting cameras on top of cars and taking pictures of all the roads.”
Jones had the difficult task of defending such programs overseas. He was like Harvey Keitel’s Cleaner character in Pulp Fiction, but instead of tidying up crime sites, his task was mediating the insults to international sensibilities caused by Google Maps and Google Earth. “I fly there not to smooth things over but to bring engineering knowledge to the debate,” he says. Some countries, such as India, prohibited mapping services on the grounds of national security. China required a license, which Google could not obtain. (That drove Google into a partnership with a legally blessed local service.) In Europe, where privacy standards are much tougher than in America, privacy commissioners did not consider pictures taken in public places appropriate for the public Internet. In talks with a surveyor general or a privacy official—or even the president of India, at one point—Jones would counter objections by explaining that Google got its geodata from public places and commercial vendors. “If it’s a security issue, you should already be scared, because we just pulled out our credit cards and bought the pictures—so certainly the bad guys could have bought the pictures, too,” he’d tell them. Of course, since Google provided the pictures, the bad guys no longer had to purchase images—Larry and Sergey’s company made them available instantly, for free. When that was pointed out to Jones, he would revert to the oft-invoked claim that every valuable technology has a potential for misuse.
Actually, the argument that Google was using only public information was no longer true. Google increasingly added its own sources of data to those it bought or accessed. In 2006, it introduced a system whereby users could annotate maps that were missing geographic data. (This feature was especially helpful in developing countries, where maps did not reflect back roads and newly cleared plots of land.) In 2009, Google combined information it gathered itself with terabytes of data bought from government and regional databases to create its own competitor to the big mapping providers Navteq and Tele Atlas. (Google had been concerned when Nokia bought Navteq for $8.1 billion in 2007 and that same year TomTom paid $4 billion for Tele Atlas, which had been providing Google with geodata.) And then there was Street View, which enabled digital drive-bys to anyone with a browser.
Google’s designers thought that the program would be universally embraced; besides admiring their own homes, people would be able to identify destinations in advance. You would be able to identify a new hair salon or restaurant, or the location of a dinner party before you left your driveway, saving time and anxiety. Or you could simply do some sightseeing at a remote location from the comfort of your LCD screen. But privacy advocates who saw previews of the product were horrified. “They would say, ‘Oh my God, that’s terrible—you can actually see a person in Times Square!’” says Jones. He thought the objection was ridiculous. So what if you’re standing in Times Square? As far as he was concerned, people who stepped out in public had implicitly given permission for people to look at them—and, by extension, for Google Street View to capture their images in the course of documenting the physical world. Still, some objections were difficult to shrug off. What about someone who was walking into a strip club—or simply walking past a strip club? What about a married person walking hand in hand with someone other than his or her spouse? What if Google StreetView showed—as spotters actually discovered—teenagers sunbathing in skimpy attire, hoodlums breaking into buildings, high school girls catfighting, and people mysteriously wearing horse heads? Did Google really want to become a global snoop?
Critics also lambasted Google for working on the project for more than a year before making it public less than eight months before its planned implementation. But Jones insists that Google’s do-first-apologize-later ethic, here and elsewhere, was essential to the company’s success. Ideas, he explained, were like babies—everything about their environment said they shouldn’t exist. But they do. You can’t dwell on problems too early, or they will swamp the virtues and you will decide not to do the project. That’s why Google managed to do so many things when other companies held back. Google understood as well as other companies that there are millions of reasons not to do something. “We keep our mouths shut about it,” he says. While the Street View project was in incubation, Google ignored the privacy issues.
Instead, Google concentrated on engineering problems. The team drove a car around Mountain View, then around the Bay Area, each time tweaking the technology. Then it adapted more cars, each time improving the ability to capture images, link them together, anchor them to geographic coordinates. Only after it went through three kinds of cameras, four kinds of GPS devices, and four separate iterations of the system itself did the team submit the project for approval. It was no longer a baby.
Over the first year of Street View, Google belatedly installed the privacy features that critics demanded. The revamped version could algorithmically detect faces and license plates, blurring them so they would not be identified. (Sometimes the algorithm was overly ambitious. “There are horses where the horses’ faces are blurred out and stuff like that,” says Jones.) In addition, Google allowed people to demand redactions if the photos made them identifiable. Google would comply, no questions asked.
But there was no way that Google would stop Street View altogether, as some critics demanded. The project was a key component in the company’s bigger information picture. What’s more, Microsoft had its own mirror world, its own fleet of camera-equipped cars cruising the streets, its own low-flying air force to capture three-quarters views of buildings for a SimCity-style picture of the real world. But Google, the market leader, got the attention—and the traffic.
But when something went wrong, the reaction was explosive. In early 2010, Google made a horrifying discovery: the cars driving around the streets of the world taking pictures for Street View had “unintentionally” sucked up confidential information—known as “payload data”—from wireless Internet transmitters in the areas they cruised. Any Wi-Fi device not protected by passwords seemed to have been vulnerable. It appeared to be a form of surveillance, snatching whatever information people were sending and receiving from the net during the brief time when the cars were passing.
After several weeks of war room analysis and furious fire dousing by its communications staff, Google presented the situation as a regrettable miscue, claiming that the problem had arisen from code an engineer had written for an experimental Wi-Fi project. The engineer’s program, Google said, “sampled all categories of publicly broadcast Wi-Fi data.” (That meant even private information on networks not protected by passwords, and in fact Google wound up collecting people’s emails, financial information, and other personal information.) Apparently, the engineers working on the Wi-Fi Street View project noticed that someone had written useful code and implemented it without understanding its intrusive nature. One Google lawyer later referred to the original engineer as a “rogue”; in any case, he was operating without instructions from any manager or director. Just as Googlers do all the time.
While the Street View team was creating a system to log the active Wi-Fi networks in the areas it mapped (to increase the accuracy of its data), it made use of that rogue code, presumably unaware that it would enable the Street View vehicles to perform surveillance activities. The mystery was why no one at Google noticed that Street View servers were loaded with gigabytes of data that had no
business being there. In any case, collecting the information was a potential violation of data security laws, and the transgression triggered investigations in several countries and states.
The incident exposed the risks that arise when tolerance of a company’s information retention policies is at the limit. Even its tiniest mistakes called attention to the larger truth—that Google had a frightening amount of information under its control. And when something major went wrong, like the Street View Wi-Fi debacle, it eroded Google’s main line of defense when justifying its stewardship of the world’s information: trust.
Google’s next antitrust crisis after DoubleClick began in February 2008, as a consequence of a hostile bid made by Microsoft to take over Yahoo. Microsoft’s $48 billion offer included an aggressive 62 percent premium over the struggling target’s share price, and so observers assumed that the merger was sealed. But Yahoo’s chairman, Jerry Yang, resisted, and his efforts to thwart the takeover were aided by Google. Within days of the offer, Eric Schmidt called Yang and began talking about a partnership that would help the weaker company. Google also began contacting legislators and regulators about the antitrust implications of the Microsoft deal, a rather odd stance considering Google’s previous insistence that the search marketplace had no lock-in and thus wasn’t a valid candidate for antitrust action. Later that spring Google took a more concrete step, hammering out an arrangement whereby some of Yahoo’s search customers would be served Google’s ads. Since Google’s ad system was much more productive, this would result in bigger profits for Yahoo, and its shareholders would presumably be more sanguine about the missed opportunity to cash in on the Microsoft deal.
Brin explained at a TGIF that in addition to the obvious desire to keep its two top rivals from combining forces, there was a personal motivation for the arrangement. “It was tough for me and Larry to turn away Yahoo,” Brin said. “They encouraged us to start this company.” (Of course, if Yang and David Filo had bought Google instead of licensing it and helping it dominate the search field, Yahoo might not have been in its current predicament.)
Google felt that it had cleverly dealt with the threat of its two closest competitors joining forces. But the search giant’s deal with the number two search engine put it back into the sights of the Department of Justice. Now it was Microsoft prodding the DOJ to look into search. Microsoft’s deep and hard-won experience with the DOJ made it a much more skillful player than the kids from Mountain View.
Microsoft began a regular series of briefings in D.C. described in the press as “screw Google” meetings. A spokesperson resisted that appellation but conceded that Microsoft was working to “educate policy makers and regulators” about the competitive landscape. Back in the 1990s, Microsoft’s corporate psyche had been brutally battered by its antitrust ordeal, with Bill Gates himself humiliated by a painful deposition that could have been worse only if there were a YouTube at the time to expose the video on an even wider scale. It used to go without saying that Gates wouldn’t wish such an experience on his worst enemy. Now Gates was doing just that. Of course, he would no doubt resist the comparison between his company’s undeserved antitrust debacle and Google’s just deserts. Google lawyers would agree that the two cases were dramatically different, but they would contend that Microsoft had been unlawfully anticompetitive in a way Google had never been.
Microsoft hired the heavyweight firm Cadwalader, Wickersham & Taft to push its anti-Google agenda to the DOJ. The key was whether the department, in the waning days of the Bush administration, would budge from its lax antitrust policy. It could go either way, depending on whether Justice accepted Google’s argument that search advertising was only a part of the galaxy of online ads or even the entire universe of advertising. There was a troubling early indicator when the questions the DOJ formally submitted to Google seemed to focus skeptically on that question and the terms of the Yahoo deal. And in September there came a development that was genuinely alarming from Google’s perspective. The DOJ contracted outside help, in the form of Sanford “Sandy” Litvack. The Chicago-based attorney was not a reflective academic schooled in subtleties of the law as much as he was a sharp-elbowed antitrust litigator. “When they say, ‘We’re going to a courtroom attorney to help,’ it’s not a happy day,” noted Google’s counsel Dana Wagner.
Indeed, not only did Litvack take a dim view of Google’s attempts to partner with Yahoo, but he prepared a broader complaint against the company. On the morning of November 5, 2008, the DOJ informed Google that later in the day it would charge the company with a violation of Section 1 of the Sherman Antitrust Act, calling the Yahoo agreement a restraint of free trade. Worse, the complaint would also accuse Google of violating Section 2 of the act, an illegal attempt to monopolize. Clearly, Litvack didn’t accept Google’s invitation to view its business as a small percentage of the advertising world. Instead he saw the company as the 80 percent dominator of search ads, the venue that every advertiser was forced to patronize. “We would have ended up also alleging that Google had a monopoly and that [the Yahoo deal] would have furthered their monopoly,” Litvack later explained to American Law Daily.
Google ruled a monopoly? The company could not let that stand. “I really did believe it was possible for us to structure an arm’s-length deal that met the antitrust terms,” Schmidt later said. “I tried hard. I talked to Sandy. It was an example where we’re running against other people’s agendas and their worldviews.” Google quickly terminated its agreement with Yahoo, informing the government only three hours before the feds filed a complaint that would have made “monopolist” a keyword when people searched for Google’s company information.
With no agreement to rule on, the government stopped its investigation. Google might have dodged a bullet but thereafter had to face the fact that the antitrust gun was loaded and pointed straight at Mountain View.
(The gambit looked even worse the next winter. With nowhere else to turn—and the economic downturn making the company a less attractive takeover target—Yahoo’s new CEO, former Autodesk head Carol Bartz, arranged to turn over Yahoo’s search business to Microsoft for a bargain price of a billion dollars. Microsoft got the main prize it had sought in the merger for barely 3 percent of its original offer.)
Google had hopes that the gun might be holstered when Barack Obama took office. “I really think this is going to be the first Internet administration,” said Google lobbyist Pablo Chavez soon after Obama’s election. Of course the new president couldn’t intervene in Google’s behalf in a legal case—but somehow the fact that Google and Obama vibrated at the same frequency seemed to portend well. “I spent a fair amount of time with him during the campaign,” said Schmidt of the new president. “He certainly understands what Google searches are, he understands our advertising model, he understands the structure of the company. He’s clearly a Google user.”
But another shock disabused Google of its hopes. Back in the spring of 2008, Google’s lawyers had been too concerned with the Yahoo agreement to notice some remarks made at an unheralded panel discussion hosted by the American Antitrust Institute. One of the speakers was Christine Varney, who had worked on Netscape’s behalf during the 1990s, successfully pushing the government to file its antitrust suit against Microsoft. Now she described Microsoft as “so last century.” The current problem was Google, which “has acquired a monopoly in Internet marketing.” Though the company may have obtained its dominance lawfully, she continued, Google is “quickly gathering market power for what I could call an online computing environment in the clouds. When all our enterprises move to computing in the clouds and there is a single firm that is offering a comprehensive solution, you are going to see the same repeat of Microsoft.”
Those words didn’t matter when they came from just another lawyer in the high-tech policy community. But in February 2009, President Obama appointed Varney as the head of the DOJ’s Antitrust Division. Suddenly the safety was off, the loaded gun pointed at Mountain View.
Virtually every expansion Google attempted from that point on—every acquisition and deal, every expansion into new territory—would require a painstaking government review, risking another decision like Sandy Litvack’s. Google even found itself fending off a theory that by virtue of having such a huge search market share, Google’s algorithmic decisions in determining search results should be subject to government approval to ensure that the company did not play favorites. Google tried to launch an offensive, including a tour of media outlets, government offices, and legislative confabs where Dana Wagner would present a slide show. (Opponents called it “the we-are-not-evil road show.”) At any turn, the Department of Justice might invite itself to weigh in on Google’s activities. And that included the project that, of all Google’s post-basic web searches, might have been the one closest to its heart.
3
“That horrid Google on the prowl!!!”
Larry Page and Sergey Brin were not literary types; they breathed Internet air, not the musty aura of paper and printer’s ink. (“Why don’t you just write some articles?” Brin asked me after an interview. “Or release this a chapter at a time?”) But they did understand the value locked inside the approximately 33 million book titles printed since Johannes Gutenberg invented the printing press. Even before Google was Google, in fact, Larry Page had been thinking about that knowledge as an adjunct to the web, humanity’s outpourings corralled into a single database that, of course, you could search. It was no coincidence that he and Brin had been affiliated with a government-funded project in grad school called the National Science Digital Library. “We tried to get book search going at Stanford,” Page would later say. “It would actually be really nice to be able to search all books,” he’d say to his professors. “Why don’t we do that?” It seemed obvious to him, but the professors deflected his suggestion. “They had other ideas about what that project was really about,” he says. “If you asked anybody about it, they would immediately decide it was impossible.”