Google's first foray into Chinese markets was a short-lived experiment. Google China’s search engine was launched in 2006 and abruptly pulled from mainland China in 2010 amid a major hack of the company and disputes over censorship of search results. But in August 2018, the investigative journalism website The Intercept reported that the company was working on a secret prototype of a new, censored Chinese search engine, called Project Dragonfly. Amid a furor from human rights activists and some Google employees, US Vice President Mike Pence called on the company to kill Dragonfly, saying it would “strengthen Communist Party censorship and compromise the privacy of Chinese customers.” In mid-December, The Intercept reported that Google had suspended its development efforts in response to complaints from the company's own privacy team, who learned about the project from the investigative website's reporting.
Observers talk as if the decision about whether to reenter the world’s largest market is up to Google: will it compromise its principles and censor search the way China wants? This misses the point—this time the Chinese government will make the decisions.
Google and China have been locked in an awkward tango for over a decade, constantly grappling over who leads and who follows. Charting that dance over the years reveals major shifts in China’s relationship with Google and all of Silicon Valley. To understand whether China will let Google back in, we must understand how Google and China got here, what incentives each party faces—and how artificial intelligence might have both of them dancing to a new tune.
The right thing to do?
When www.google.cn launched in 2006, the company had gone public only two years before. The iPhone did not yet exist, nor did any Android-based smartphones. Google was about one-fifth as large and valuable as it is today, and the Chinese internet was seen as a backwater of knockoff products that were devoid of innovation. Google’s Chinese search engine represented the most controversial experiment to date in internet diplomacy. To get into China, the young company that had defined itself by the motto “Don’t be evil” agreed to censor the search results shown to Chinese users.
Central to that decision by Google leadership was a bet that by serving the market—even with a censored product—they could broaden the horizons of Chinese users and nudge the Chinese internet toward greater openness.
At first, Google appeared to be succeeding in that mission. When Chinese users searched for censored content on google.cn, they saw a notice that some results had been removed. That public acknowledgment of internet censorship was a first among Chinese search engines, and it wasn’t popular with regulators.
“The Chinese government hated it,” says Kaiser Kuo, former head of international communications for Baidu. “They compared it to coming to my house for dinner and saying, ‘I will agree to eat the food, but I don’t like it.’” Google hadn’t asked the government for permission before implementing the notice but wasn’t ordered to remove it. The company’s global prestige and technical expertise gave it leverage. China might be a promising market, but it was still dependent on Silicon Valley for talent, funding, and knowledge. Google wanted to be in China, the thinking went, but China needed Google.
Google’s censorship disclaimer was a modest victory for transparency. Baidu and other search engines in China soon followed suit. Over the next four years, Google China fought skirmishes on multiple fronts: with the Chinese government over content restrictions, with local competitor Baidu over the quality of search results, and with its own corporate leadership in Mountain View, California, over the freedom to adapt global products for local needs. By late 2009, Google controlled more than a third of the Chinese search market—a respectable share but well below Baidu’s 58%, according to data from Analysys International.
In the end, though, it wasn’t censorship or competition that drove Google out of China. It was a far-reaching hacking attack known as Operation Aurora that targeted everything from Google’s intellectual property to the Gmail accounts of Chinese human rights activists. The attack, which Google said came from within China, pushed company leadership over the edge. On January 12, 2010, Google announced, “We have decided we are no longer willing to continue censoring our results on Google.cn, and so over the next few weeks we will be discussing with the Chinese government the basis on which we could operate an unfiltered search engine within the law, if at all.”
The sudden reversal blindsided Chinese officials. Most Chinese internet users could go about their online lives with few reminders of government controls, but the Google announcement shoved cyberattacks and censorship into the spotlight. The world’s top internet company and the government of the most populous country were now engaged in a public showdown.
“[Chinese officials] were really on their back foot, and it looked like they might cave and make some kind of accommodation,” says Kuo. “All of these people who apparently did not give much of a damn about internet censorship before were really angry about it. The whole internet was abuzz with this.”
But officials refused to cede ground. “China welcomes international Internet businesses developing services in China according to the law,” a foreign ministry spokeswoman told Reuters at the time. Government control of information was—and remains—central to Chinese Communist Party doctrine. Six months earlier, following riots in Xinjiang, the government had blocked Facebook, Twitter, and Google’s YouTube in one fell swoop, fortifying the “Great Firewall.” The government was making a bet: China and its technology sector did not need Google search to succeed.
Google soon abandoned google.cn, retreating to a Hong Kong–based search engine. In response, the Chinese government decided not to fully block services like Gmail and Google Maps, and for a while it allowed sporadic access from the mainland to the Hong Kong search engine too. The two sides settled into a tense stalemate.
Google’s leaders seemed prepared to wait it out. “I personally believe that you cannot build a modern knowledge society with that kind of [censorship],” Google chairman Eric Schmidt told Foreign Policy in 2012. “In a long enough time period, do I think that this kind of regime approach will end? I think absolutely.”
But instead of languishing under censorship, the Chinese internet sector boomed. Between 2010 and 2015, there was an explosion of new products and companies. Xiaomi, a hardware maker now worth over $40 billion, was founded in April 2010. A month earlier Meituan, a Groupon clone that turned into a juggernaut of online-to-offline services, was born; it went public in September 2018 and is now worth about $35 billion. Didi, the ride-hailing company that drove Uber out of China and is now challenging it in international markets, was founded in 2012. Chinese engineers and entrepreneurs returning from Silicon Valley, including many former Googlers, were crucial to this dynamism, bringing world-class technical and entrepreneurial chops to markets insulated from their former employers in the US. Older companies like Baidu and Alibaba also grew quickly during these years.
The Chinese government played contradictory roles in this process. It cracked down on political speech in 2013, imprisoning critics and instituting new laws against “spreading rumors” online—a one-two punch that largely suffocated political discussion on China’s once-raucous social-media sites. Yet it also launched a high-profile campaign promoting “mass entrepreneurship and mass innovation.” Government-funded startup incubators spread across the country, as did government-backed venture capital.
That confluence of forces brought results. Services like Meituan flourished. So did Tencent’s super-app WeChat, a “digital Swiss Army knife” that combines aspects of WhatsApp, PayPal, and dozens of other apps from the West. E-commerce behemoth Alibaba went public on the New York Stock Exchange in September 2014, selling $25 billion worth of shares—still the most valuable IPO in history.
Amidst this home-grown success, the Chinese government decided to break the uneasy truce with Google. In mid-2014, a few months before Alibaba’s IPO, the government blocked virtually all Google services in China, including many considered essential for international business, such as Gmail, Google Maps, and Google Scholar. “It took us by surprise, as we felt Google was one of those valuable properties [that they couldn’t afford to block],” says Charlie Smith, the pseudonymous cofounder of GreatFire, an organization that tracks and circumvents Chinese internet controls.
The Chinese government had pulled off an unexpected hat trick: locking out the Silicon Valley giants, censoring political speech, and still cultivating an internet that was controllable, profitable, and innovative.
AlphaGo your own way
With the Chinese internet blossoming and the government not backing down, Google began to search for ways back into China. It tried out less politically sensitive products—an “everything but search” strategy—but with mixed success.
In 2015, rumors swirled that Google was close to bringing its Google Play app store back to China, pending Chinese government approval—but the promised app store never materialized. This was followed by a partnership with Mobvoi, a Chinese smart-watch maker founded by an ex-Google employee, to make voice search available on Android Wear in China. Google later invested in Mobvoi, its first direct investment in China since 2010.
In March 2017, there were reports that authorities would allow Google Scholar back in. They didn’t. Reports that Google would launch a mobile-app store in China together with NetEase, a Chinese company, similarly came to naught, though Google was permitted to relaunch its smartphone translation app.
Then, in May 2017, a showdown between AlphaGo, the Go-playing program built by Google sibling company DeepMind, and Ke Jie, the world’s number one human player, was allowed to take place in Wuzhen, a tourist town outside Shanghai. AlphaGo won all three games in the match—a result that the government had perhaps foreseen. Live-streaming of the match within China was forbidden, and not only in the form of video: as the Guardian put it, “outlets were banned from covering the match live in any way, including text commentary, social media, or push notifications.” DeepMind broadcast the match outside China.
During this same period, Chinese censors quietly rolled back some of the openings that Google’s earlier China operations had catalyzed. In 2016, Chinese search engines began removing the censorship disclaimers that Google had pioneered. In 2017, the government launched a new crackdown on virtual private networks (VPNs), software widely used for circumventing censorship. Meanwhile, Chinese authorities began rolling out extensive AI-powered surveillance technologies across the country, constructing what some called a “21st-century police state” in the western region of Xinjiang, home to the country’s Muslim Uighurs.
Despite the retrograde climate, Google capped off 2017 with a major announcement: the launch of a new AI research center in Beijing. Google Cloud’s Chinese-born chief scientist, Fei-Fei Li, would oversee the new center. “The science of AI has no borders,” she wrote in the announcement of the center’s launch. “Neither do its benefits.” (Li left Google in September 2018 and returned to Stanford University, where she is a professor.)
If the research center was a public symbol of Google’s continued efforts to gain a foothold in China, Google was also working quietly to accommodate Chinese government restrictions. Dragonfly, the censored- search-engine prototype, which has been demonstrated for Chinese officials, blacklists key search terms; it would be operated as part of a joint venture with an unnamed Chinese partner. The documents The Intercept obtained said the app would still tell users when results had been censored.
Other aspects of the project are particularly troubling. Prototypes of the app reportedly link users’ searches to their mobile-phone number, opening the door to greater surveillance and possibly arrest if people search for banned material.
In a speech to the Dragonfly team, later leaked by The Intercept, Ben Gomes, Google’s head of search, explained Google’s aims. China, he said, is “arguably the most interesting market in the world today.” Google was not just trying to make money by doing business in China, he said, but was after something bigger. “We need to understand what is happening there in order to inspire us,” he said. “China will teach us things that we don’t know.”
In early December, Google CEO Sundar Pichai told a Congressional committee that "right now we have no plans to launch in China," though he would not rule out future plans. The question is, if Google wants to come back to China, does China want to let it in?
To answer that question, try thinking like an advisor to President Xi Jinping.
Bringing Google search back certainly has upsides. China’s growing number of knowledge workers need access to global news and research, and Baidu is notoriously bad at turning up relevant results from outside China. Google could serve as a valuable partner to Chinese companies looking to expand internationally, as it has demonstrated in a patent-sharing partnership with Tencent and a $550 million investment in e-commerce giant JD. Google’s reentry would also help legitimize the Communist Party’s approach to internet governance, a signal that China is an indispensable market—and an open one—as long as you “play by the rules.”
But from the Chinese government’s perspective, these potential upsides are marginal. Chinese citizens who need to access the global internet can still usually do so through VPNs (though it is getting harder). Google doesn’t need to have a business in China to help Chinese internet giants gain business abroad. And the giants of Silicon Valley have already ceased their public criticism of Chinese internet censorship, and instead extol the country’s dynamism and innovation.
By contrast, the political risks of permitting Google to return loom large to Xi and his inner circle. Hostility toward both China and Silicon Valley is high and rising in American political circles. A return to China would put Google in a political pressure cooker. What if that pressure—via antitrust action or new legislation—effectively forced the company to choose between the American and Chinese markets? Google’s sudden exit in 2010 marked a major loss of face for the Chinese government in front of its own citizens. If Chinese leaders give the green light to Project Dragonfly, they run the risk of that happening again.
A savvy advisor would be likely to think that these risks—to Xi, to the Communist Party, and to his or her own career—outweighed the modest gains to be had from allowing Google’s return. The Chinese government oversees a technology sector that is profitable, innovative, and driven largely by domestic companies—an enviable position to be in. Allowing Google back in would only diminish its leverage. Better, then, to stick with the status quo: dangle the prospect of full market access while throwing Silicon Valley companies an occasional bone by permitting peripheral services like translation.
Google does have one factor in its favor. If it first entered China during the days of desktop internet, and departed at the dawn of the mobile internet, it is now trying to reenter in the era of AI. The Chinese government places high hopes on AI as an all-purpose tool for economic activity, military power, and social governance, including surveillance. And Google and its Alphabet sibling DeepMind are the global leaders in corporate AI research.
This is probably why Google has held publicity stunts like the AlphaGo match and an AI-powered “Guess the Sketch” game on WeChat, as well as taking more substantive steps like establishing the Beijing AI lab and promoting Chinese use of TensorFlow, an artificial-intelligence software library developed by the Google Brain team. Taken together, these efforts constitute a sort of artificial-intelligence lobbying strategy designed to sway the Chinese leadership.
This pitch, however, faces problems on at least three battlegrounds: Beijing; Washington, DC; and Mountain View, California.
Chinese leaders have good reason to feel they’re already getting the best of both worlds. They can take advantage of software development tools like TensorFlow and they still have a prestigious Google research lab to train Chinese AI researchers, all without granting Google market access.
In Washington, meanwhile, American security officials are annoyed that Google is actively courting a geopolitical rival while refusing to work with the Pentagon on AI projects because its employees object to having their work used for military ends.
Those employees are the key to the third battleground. They’ve demonstrated the ability to mobilize quickly and effectively, as with the protests against US defense contracts and a walkout last November over how the company has dealt with sexual harassment. In late November more than 600 Googlers signed an open letter demanding that the company drop the Dragonfly project, writing, “We object to technologies that aid the powerful in oppressing the vulnerable.” Daunting as these challenges sound—and high as the costs of pursuing the Chinese market may be—they haven’t entirely deterred Google’s top brass. Though the development of Dragonfly appears to have, at the very least, paused, the wealth and dynamism that make China so attractive to Google also mean the decision of whether or not to do business there is no longer the company’s to make.
“I know people in Silicon Valley are really smart, and they’re really successful because they can overcome any problem they face,” says Bill Bishop, a digital-media entrepreneur with experience in both markets. “I don’t think they’ve ever faced a problem like the Chinese Communist Party.”
Matt Sheehan is a fellow at MacroPolo and worked with Kai-Fu Lee on his book AI Superpowers.
The big new idea for making self-driving cars that can go anywhere
The mainstream approach to driverless cars is slow and difficult. These startups think going all-in on AI will get there faster.
Inside Charm Industrial’s big bet on corn stalks for carbon removal
The startup used plant matter and bio-oil to sequester thousands of tons of carbon. The question now is how reliable, scalable, and economical this approach will prove.
The hype around DeepMind’s new AI model misses what’s actually cool about it
Some worry that the chatter about these tools is doing the whole field a disservice.
The dark secret behind those cute AI-generated animal images
Google Brain has revealed its own image-making AI, called Imagen. But don't expect to see anything that isn't wholesome.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.