A dream deferred: The state of the Chinese Internet, June 4, 2009.
Tiananmen 20 years on
Laogai Research Foundation/NED
Panel 1: Refinement of Repression, 9:15 am, June 2, 2009
Ethan Gutmann: “The Chinese Internet: A dream deferred?”
Of all the post-Tiananmen dreams of liberation, the case of the Chinese Internet is among the most tragic—on a number of levels—and I’ll be speaking about that in broad terms today. Yet subsequent attempts to liberate China’s Big Brother Internet have explosive potential and I’ll talk a little about that as well.
After Tiananmen, a frontal attack on the party’s hold on the Chinese State appeared impossible, and attention shifted to a very different strategy opened up by Deng’s Southern Tour. Perhaps a great wave of entrepreneurial activity, technology and commercialism could lead the forces of reform, and simply sweep away the rotting system. Thus, as the growth of the Internet, and the so-called “New Economy”—fuelled by the infusion of western capital—appeared at the end of the Nineties, it seemed an answer to a prayer, offering a sort of technological end-run around the Party system and the horrors of China’s recent past.
A decade ago, the Internet was small, but it was also in a “state of nature.” The Party blocked sites, but it was a clumsy, slow affair and easily circumnavigated; the most common search terms in China were not “Britney” and “Hooters” but “proxy” and “server”—a method to surf the free web, and cloak one’s identity at the same time. Dissident messages were even considered moderately “safe for work.” Rumors, critiques of the Party, gossip—often from a nationalist or a labor perspective would continuously flare up on my co-workers screens, while Internet cafes—the real ones, the cheap ones that didn’t serve lattes—were packed tight with computers and young Chinese. They sprung up in the hutongs, like crack-houses in an American ghetto, right next to the open-air toilets.
Perhaps because of this fleeting moment, Western cyber-utopians have never understood the Party’s relationship with the Internet. It bears repeating that they are Marxists. That means: Embrace the means of communication. Control it. Use it. Fill it with Chinese voices. Prevent dissident groups from using it as a platform or a space to end their isolation. Instead enforce their isolation. And when they try to break out, arrest them.
At the same time, the Party had learned from near-thing-Tiananmen: don’t put your faith in the big battalions because they can flip. So diversify. Fill Beijing with small military units and enough will stay loyal, but make sure the avenues can sustain the weight of armored vehicles as they rush to the center. So let the Internet Service Providers proliferate. Let the Westerners in. But make sure the main Internet corridors are patrolled by the Party.
They called this approach the Gold Shield. The monolithic sound of that is a little misleading because there were two different methods to accomplish these goals. The first, and the one that gets all the attention, was to censor the web. Now contrary to Western naval-gazing the censorship wasn’t aimed at CNN.com (unless there was a particularly sensitive story on China). It was aimed at labor sites, Maoist sites, overseas dissident sites, Taiwanese, Falun Gong, and Chinese Christians.
Blocking the outside was relatively easy, but the Party needed help particularly for the internal sites. Three companies were competing for the Chinanet contracts in 1997: Bay Networks, Sun Microsystems, and Cisco Systems. Cisco prevailed by selling the authorities a “firewall box” at a significant discount, which would allow the Chinese authorities to comprehensively block the forbidden web.
Cisco’s General Counsel denies selling any special configuration. Chinese engineers who actually worked on the firewall project are equally adamant that it was custom-made. Either way, as early as late 1998, any industry-wide restraints on the transfer of censorship technologies were already being weighed against Cisco’s capture of 80% of the China router market, an unprecedented success story.
By 2000, Yahoo began censoring its search engine and patrolling chatrooms to preserve its position as the top portal in China. According to Yahoo’s former China manager: “It was a precautionary measure. The State Information Bureau was in charge of watching and making sure that we complied. The game is to make sure that they don’t complain.” Eventually Google (and Microsoft, to a lesser extent) would avoid complaints using Yahoo’s method.
Yahoo would eventually get caught assisting the arrest of journalist Shi Tao, but in terms of raw numbers, Yahoo was actually a bit player. Here’s why: The second Gold Shield method was surveillance. And that’s far more technically challenging, and far more important. If you can watch people, and they know you are watching, they will censor themselves (in the case of Chinese blogs, you will even observe them displaying a certain pride in towing the Party line). And surveillance means you can catch them. So in 2001, Internet Service Providers (ISPs) were told to record all user data—surfing history, email, possibly even keystrokes—for 60 days. But that wasn’t enough. Computers were usually shared among many different users. And proxies were being hunted, but they kept popping up. What was needed was wraparound surveillance, a way to attach the individual surfer’s online presence to other communications and, ultimately, to their danwei—their employment and political history file.
Now the Party didn’t have the technical capability to do this. In fact it wasn’t clear that anyone really did (at least on a large scale). Nortel believed surfer identities could be tied to facial recognition, a very technically demanding—and expensive—strategy. Trial runs produced several arrests of dissidents in train stations and other locations where surveillance cameras were present, but it also created way too many false positives.
Perhaps identity should be tied to fingerprint recognition? A Chinese company named Goldenfinger, pushed this idea, but, as the name suggests, it was clearly too literal, too clumsy. Proximity readers? Sun Microsystems expressed some interest in this idea and Xerox was rumored to be hunting around for a contract to produce a national ID card with a chip, but Cisco Systems had a better idea. They proposed that identity should be based on the one thing that everyone in China would soon have—a cell phone. By tying voice patterns to surf patterns and then to work and political history patterns (the danwei)—and then, making that file accessible—mobile accessible—to local police to the PSB to the courts—well, you had a system that allowed a policeman to stop a Chinese citizen on the street and begin reading their email or what their elementary school teacher had to say about their ideological leanings. A paperless legal system.
But this isn’t science fiction—these are based on Cisco’s own documents. Documents that explicitly show Cisco was training the Chinese police in surveillance techniques as early as 2001, that Cisco technology should be used to combat “Falun Gong evil cult and other hostile elements,” and so on. The police had compiled a vast database of Falun Gong practitioners by 2001. Individual locations could be tracked in real time using cell phones. In fact, mobile phones were probably responsible for more arrests—and thus imprisonment and murder–of Falun Gong, and other dissidents, than any other method.
Now there were other Western companies involved in these activities, but I think it’s fair to say that Cisco—and it’s avoidance of any serious consequences—created the industry template. And Cisco has never denied that it has contracts with Chinese state security, preferring to capitalize on the general confusion between censorship—where they have plausible deniability—and surveillance—where the evidence is indisputable. But by the time the first Global Online Freedom Act was drafted in 2006—by the time Congress began to think seriously about the fact that there are explicit laws against selling surveillance assets to the Chinese police, laws passed in the wake of Tiananmen—it was increasingly irrelevant.
The Party never trusted foreign companies. Go back and examine the Party’s narrowly averted attempt to make Microsoft’s encryption illegal in 2000. And let’s give Chinese homegrown ingenuity its due. For example, following Falun Gong’s successful hijacking of television signals in Changchun in early 2002, the local cable company came up with a method to monitor whether any individuals were watching politically illicit DVDs in the Changchun area. Cisco was a means to an end, a way to incorporate advanced technology (and incidentally to make Huawei stronger).
The West was semi-aware of these events. Western whiz-kids with handles like Triangle Boy and Peekabooty rushed in to free the Chinese Internet. Some even got a smattering of government funding. But their elegant theoretical hacks quickly collapsed under the brute size of the Chinese Internet Spying Agency. Washington was always looking for a magic bullet, a single-algorithm solution if you will. There was none. And the dream should end there, with Chinese free expression drying up like a raisin in the sun.
Yet Dynaweb was different. Run out of a suburban home in North Carolina, the tiny company was manned by Chinese Falun Gong practitioners whose engineering training dated back to the early days of the Chinese web. Because the Chinese Internet was constructed in Levittown fashion, all at once, it was configured in blocks. Dynaweb engineers, by looking at a seemingly meaningless series of numbers, could tell with reasonable certainty if an address was state-connected. With this knowledge and powerful hacking skills, they could track state security behavior as if they were watching a dog taking his daily morning stroll. Eventually they saw patterns; corners where the dog gave only a cursory sniff. Dynaweb built special programs simulating those corners, but with hairline cracks in the firewalls. Then Dynaweb drove truck after truck through them, loaded with uncensored news, the Nine Commentaries—Falun Gong’s Ur-condemnation of the Party—and instructions for Chinese users to set up their own wormholes.
It was a dynamic process. Bill Xia, the public face of the company remembers waking up one morning to find that the Party censors had disabled Dynaweb. He corrected the system. Ten minutes later, it was down again. And so it went for months, like salvos between the trenches, with activity falling into a lull on the Chinese Spring Festival, then starting up again after the Party guys got back in the office. But Dynaweb had began working on new methods: spam attacks on an unseen scale, dummy websites that would proliferate automatically, and evolutionary algorithms that would change Internet addresses within a single second yet would somehow bookmark normally. These techniques caused havoc for the Party censors—cases of fratricide where entire Chinese government news-sites were blocked because they contained similar language to Dynaweb’s, days where the entire Chinese Internet slowed to a crawl with incalculable losses to China’s ecommerce. Dynaweb was waging a war of mobility, and they were soon joined by diverse reinforcements: Ultrareach, Freenet, and the other dissident systems that today make up the Global Internet Freedom Consortium.
The consortium became a lifeline for the underground network of Chinese practitioners and a conduit for uncensored information and dissident writings, metastasizing from the Chinese Internet into printouts and pamphlets which were dropped off on doorsteps across the villages of China—usually in the dead of night with a barking dog serving as the soundtrack.
In response, Chinese citizens began clicking a button that renounced the Chinese Communist Party (not necessarily to “quit” the Communist Party as Falun Gong likes to say—most of these people probably weren’t Party members). Yet for all the Falun Gong hype and (understandable) wishful thinking, the significance is very real. These were promissory gestures of rejection from Chinese citizens of all backgrounds and beliefs. And while the numbers are as shaky as any Internet-based survey, I think we can say with confidence that it is in the millions.
The Consortium is allowing well over half a million Chinese citizens a day to surf freely. When it became apparent that not just boring old malcontent Chinese, but young Iranians were using the Consortium systems as well, the New York Times even deigned to report on this phenomenon.
Yet we have to measure these achievements against the impressive successes on the Party side. The Chinese Internet has seen spectacular growth in all areas. And it’s like the matrix. It looks real. Blogging flourishes. News circulates, some of it from Party-sympathetic Western reporters and Phoenix TV. In most countries, the revelation of the tofu-construction schoolhouses in Szechuan would have simply brought the government down. In China, reporters were so gratified to be given a free hand to report on dramatic rescues for two weeks that when they were told to clear out and ignore the wailing parents, they actually did. Internet discussion boards appear to be free and unfettered (although the inevitable patriot comments are reputedly paid at 50 cents an entry). But that’s the point. This is the real “new economy”–the Party has turned censorship and surveillance into a viable economic model, even a profitable one.
So in conclusion: will this dream deferred, as Langston Hughes put it, just sag like a heavy load? Or will it explode?
To think about that, here are three crude models:
- Party victory. The censorship/surveillance model spreads globally, becoming increasingly profitable and accepted. Watch for the US to relinquish control of the Internet to an International body.
- Tiananmen revival, but with a better communication tool than the fax machine. China falls into crisis. Watch for a massive slowdown of the Internet at a critical moment (in part through technologies the Dynaweb has developed), followed by a virtual tipping point where suddenly no-one cares about surveillance.
- Stalemate. On a limited budget, the Consortium will keep a certain amount of free surfing going, and the Party will have to spend 100 times as much to suppress it, but the free Internet will never achieve the critical mass necessary to become an organized tool of dissent. And a tie goes to the Party.
Now, ultimately the Internet is just a tool, just one element. The real outcome of these scenarios hinges on things that I haven’t addressed in my talk today: the fractured quality of overseas Chinese dissent for example, the ability of the Chinese intellectual class to look beyond their immediate self interest, and the apparent inability of American corporations, and increasingly the US government, to imagine a profitable China without the Party.