The Google Guys (16 page)

And at the most recent annual meeting in 2008, Sergey abstained from voting against the latest proposals to do something about Chinese censorship, proposals that would setup new committees to examine Google's censorship policies in China. He felt there was some merit to the ideas. But neither did he vote in favor of the proposals.

Sergey also clings to the belief that Google can make a difference in China by pushing the envelope of how to censor. When Google started its
Google.cn
site, for example, it took the extra step of letting searchers know that information they were looking for had been removed in accordance with government laws. “Even if we can't deliver that information, at least a user knows it's gone, that there was information that was blocked,” says Wong. “That's a form of transparency Chinese users had never seen before. We're never going to be happy about removing information, but if we can make people aware of it, it's still moving the ball forward.”

Since Google started that practice and got away with it, the other Chinese search engines have started posting similar notices, and it has become standard practice.

Wong also argues that foreign companies doing business in China, including Google, can't help but make Western attitudes rub off on the country's people. “We are employing some of the smartest computer scientists in China. They are working on hard problems and [are being exposed to] some of our culture and our mission as they do so. That's something that we will bring to China that I feel really good about.”

And with effort, sometimes questionable items do get past the government censors. The lists provided by the Chinese censors are not always consistent, and leave room for interpretation. A study from the University of Toronto in 2008 found that not all search engines censor equally. When researchers tested a series of controversial subjects on the major search engines in China, they found that Baidu had filtered out 26.4 percent of the sites, Yahoo censored 20.8 percent, while MSN and Google were virtually tied, at 15.7 and 15.2 percent, respectively. An earlier study by Reporters Without Borders found similar results. While Google fared the best, the researchers at the University of Toronto's Citizen's Lab found that all of the search engines could be doing better. The lab found that 313 sites were censored by at least one of the search engines, but only 76 were censored by all four. Perhaps this leaves room for creating an independent committee at Google to examine more carefully what it censors.

Still, Google's management insists they offer tougher competition to which competitors will have to respond. Again, this does not apply just to China. “We have the same process for all of our other domains, even in the U.S.,” says Wong. “Most of the removals in the U.S. are related to copyright issues. In those cases, we also put up a notice” that the item has been removed.

Like it or not, Google, with its powerful presence on the Internet, is becoming the world's censor. As a private company, it has the right to censor from its sites whatever it deems objectionable. (Sergey refuses to accept ads for cigarettes or alcoholic beverages.) It's unquestionably a heady responsibility for any company or pair of individuals, and nobody can be faulted for worrying over how far Google will take it.

It 's not a matter of whether or not someone's watching over you. It's just a question of their intentions.

 

 

T
he other important decision Google's top three executives argue about behind closed doors is how to keep their users' personal information confidential. People want, and deserve, access to information about government officials and other public figures, but everyone wants their own information kept from prying eyes. On this issue, Larry and Sergey come down like a falling rock on the side of confidentiality. They face vigorous criticism from outside advocates.

Google collects an enormous amount of information about the people who use its services, perhaps more than any other company in history. Google's computers track what ads users click on, what items they search, what sites they go to, even the topics they write about in e-mails. The information is used to enable Google's computers to figure out how to deliver more relevant data—and ads—to every individual. It's not all about PageRank anymore.

This doesn't mean someone at Google is reading people's e-mails. Hundreds of millions of people use Google's services, and humans can't possibly be up to that task. It's all done by Google's computers without human intervention. But the data is there on Google's computers, vulnerable to government subpoenas and, perhaps, to clever hackers intent on thievery.

To Larry and Sergey, collecting such data is all part of the process of making Google's services useful. But people are suspicious of their intent. Google could, for example, sell information to advertisers and other companies, just as magazines sell information about their readers to others. There is no evidence Google has ever done this, and Larry and Sergey promise the company never will—at least not as long as they're in charge.

Explains Larry: “Whenever you do a search, you're trusting us to give you the right things. We take that very seriously. We have a pretty good reputation in that regard. I think people see us as willing to take positions that some might find weird initially. But we definitely can explain why we did it, and we're up front about that.”
¹

But if any government hits Google with a legal subpoena, the company has to turn over the data requested. Technically, such data do not identify an individual, but merely the Internet address of a user's computer. But the data can be used to find individuals—or at least those whose computers were used—by digging up the owners of the computer at that address through other means.

Privacy advocates and certain governments press Google to delete this information after a short period of time. Governments do so naively, believing that the potential threat is from Google's abusing the information; but it actually means it will make it harder for the government to snoop out illegal activity. The dispute Larry and Sergey have with deleting data is how long that period should be.

Wong says it comes down to one other reason Google collects user data: it helps determine traffic patterns that may identify cybercrooks, enabling the company to prevent similar attacks in the future. In this case, Wong again argues that preserving the data is the best way to serve users. “Focusing on users is not just delivering the best and most robust service, but also delivering a service that the user will trust. A component of that trust is about privacy because this is such a data-driven service.”

User trust is an absolute necessity for Google, and violating it—or even the perception of violating it—threatens the company's future competitiveness. It's also one of Wong's responsibilities. As part of her job, she meets regularly with product developers, starting at the early design stage, when a new product is still just a sketch on a whiteboard. She asks the developers what information they plan to collect with the product, how they plan to use it, with whom the data will be shared, and how it will be kept secure. The issues are then discussed with Larry, Sergey, and Eric, and their recommendations are designed into the product.

One of the mandates is to inform users rigorously about how the information is used. When someone downloads Google's toolbar, they have to click on a box in order to allow PageRank to collect data about their surfing habits. If they do so, a privacy notice pops up to explain how the information is used, with a heading in bold red letters that says: “PLEASE READ CAREFULLY: IT'S NOT THE USUAL YADA YADA.”

Larry and Sergey set the mandate that users must be provided with transparency—a clear explanation of what the company is up to—and the choice of whether or not to allow it. Even if the users decide they don't want their data collected, they can still download the toolbar without activating PageRank. That bucks the trend still prevalent for many Internet companies that says, essentially, “If you don't accept the terms then you can't use the product.”

Similarly, when using Google's instant-messaging system, Google Talk, users can choose the option of going off the record, which prevents Google—and the people chatting—from keeping any record of the conversation.

Still, most people do not choose the privacy options, and Google keeps the data. But it has learned to compromise. Originally, Google did not put any limitation on the length of time it would keep the data. Privacy advocates complained, both to Google and to government regulators. So, in 2007, Wong began a series of meetings with the product designers, asking them why they needed such data. The logs improve searches, spell-check features, and other services. But a big part of the reason they keep data, said the developers, was to ensure the security of the networks.

Any Internet company is subject to spam, fraud, and attacks known as “denial of service,” in which a site is inundated with dozens of automatically generated requests in order to slow it down or make it crash. By keeping the data, Google can identify any patterns that led up to an attack and the computers used to instigate it, using the information to prevent similar future attacks.

“The fact of the matter is that the person successfully attacking us today has probably been trying for two years,” says Wong. “So when we go back into the logs for a substantial amount of time, we're able to detect the pattern we have today. We can figure out all of the patterns we're seeing in an attack. We ask, ‘What's the next step of this attack? What's the best way for us to try and stop it?' It is a historical record that helps us get to the answer that we need today.”

Wong then asked the developers how long they actually needed to keep the data in order to maintain security. After collecting that feedback and reporting to Larry and Sergey, Google set its policy: it would keep the data for only eighteen months, “anonymizing” it after that so no individual or computer could be identified.

This placated no one. So Google engineers kept working on the problem to see if they could cut that retention time. The end result was that in 2008, Google announced that it would reduce the length of time data were kept by half, to nine months. “Our engineers said that they thought they could still get pretty good results, pretty good robustness, pretty good security, based on nine months,” says Wong. “And I'll be really candid, we're giving up some quality, some ability to get really good search results faster, because we don't have that historical record. But we think we've built a system now, and improved our analytical tools, that will get us what we really, really need in nine months.”

This presented another dilemma that forced Larry and Sergey to compromise. To them, user experience is of paramount importance, and the complaints about privacy were overblown. There has never been a documented case of Google violating its users' privacy, either losing control accidentally or engaging in such practices as selling information to spammers. Every complaint about Google falls into the category of what could happen, not what has happened. In general, Google has done a better job of protecting people's privacy than its competitors. In August 2005, the U.S. Department of Justice issued subpoenas to Google, America Online, Microsoft, and Yahoo to turn over two months' worth of search queries and all the URLs in their indexes, to aid in the Bush administration's defense of an Internet pornography law. Every company except Google quickly complied with the subpoenas. Google challenged it in court as a fishing expedition with no probable cause. In the end, Google managed to restrict the information to fifty thousand Web addresses, and did not have to turn over information on the key words users were using to search.

In fact, most Internet users worry about privacy less than they should and do little to protect themselves from it. But they do want the companies they deal with to provide the protection for them, and they rely on privacy advocates to force the issue on their behalf. The privacy advocates and government regulators raise the complaints with input from an extreme minority of users. This means that by cutting the data-retention time, Larry and Sergey accepted a compromise they feel is not really necessary, and sacrificed some of the user experience as a result, which can be seen as a violation of Google's promise to focus on the needs of its users.

Wong says that acceptance comes from the need to maintain the public trust. The public complaints from privacy advocates affect the overall public view of Google. “It's a tough balance,” she says. “But if we don't do something to respect our users' privacy, that can be just as deadly as delivering a poor service. If the public thinks we're selfishly keeping data, that we're at risk of a hacking attack and all of this data is stored there, then we're going to lose their trust and they won't come no matter how good the service is.”

That may still be a problem. Many privacy advocates, including the well-respected Electronic Frontier Foundation (EFF) and the European Union, still want Google to cut the retention time to six months. Wong says it's not possible without seriously compromising security and user experience.

Nevertheless, Wong says the company works well with organizations such as the EFF, where her point person is one of her former colleagues. All their recommendations are taken to the team, including Larry and Sergey, for discussion.

But a large number of complaints also come from people who express their opinions in the press without ever having spoken with Google or tried the products in question. Wong says she gets her fair share of complaints from people who don't really seem to want to solve problems, but “just want to cause a harangue.” Some of the complaints are just crazy, and are dismissed with anger. One advocate, for example, called her and argued that Google should not retain any users' search histories at all, even though users frequently use them to find sites they've visited in the past that they want to visit again. The advocate's suggestion: users should just write down on paper the URLs of sites they've visited, keeping them safe from snooping eyes, or at least those who don't have access to their desk drawers.