Ever since China linked up to the web in 1994, its rulers have sought to know, control and limit what their citizens read and write online. In the early years, the censorship system they built became known as the “Great Firewall of China”, because it focused on using router technology to block unwanted information from outside at the point where it might enter.

But as internet use has grown – China overtook the US last year as the nation with the largest number of users, an estimated 350 million according to latest figures – so too has the number of censors. And as China’s presence on the web has developed, with a greater focus on user-generated content, so have the censors’ strategies evolved. Over the past few years people such as Wu have created a multi­layered operation that watches the internet around the clock, reports the results to the leadership and engineers ever more sophisticated responses to dissent – organising the sorts of “fact-finding missions” that calmed down netizens concerned about the death of Li. “Public opinion on the internet must be solved with the means of the internet,” Wu told me. And then he went on to explain how.

. . .

Like the entire political apparatus in China, the censorship machine is controlled by two institutional bodies: the Communist party and the government. At the national level, the propaganda department of the party and the information office of the state council (the cabinet) are in charge. But these institutions only deal with big, strategic issues or nationwide challenges to the party’s image and power.

They decided which propaganda campaigns to run ahead of the 60th anniversary of the foundation of the People’s Republic of China, for example; they also issued orders to the media last month to stop reporting criticism of Green Dam Youth Escort, a content-control software that the government ordered to be installed on all new computers from July 1 (a deadline that has subsequently been pushed back).

Day-to-day surveillance and control of the population are carried out by a far greater number of departments: the double structure of censorship institutions is duplicated at the provincial, county and city level; in addition, every government department operates its own internet surveillance. “Every ministry has special departments for collecting and surveying information from the internet,” said Wu, “including the police, the telecom departments, the departments for foreign affairs and the development and research commission [the top economic policy planning body].” Together, the authorities keep a 24-hour watch on what is said online.

As for particular duties, the police – 1.9 million strong, under the control of the ministry of public security – work the frontlines. Although they refuse to comment even on whether they supervise the internet, an insight into their operations comes from those who supply them with the technical wherewithal. “Currently [the police force] still does surveillance via keyword searches on search engines, with every officer being given a certain number of keywords to cover,” says a marketing manager at Beijing TRS Information Technology. Increasingly, however, more advanced methods are being employed, such as the use of “data-mining” software. “We equipped eight police stations in Shanghai with such equipment,” says the manager. “Now the work of 10 internet cops can be done by just one.”

The “internet cops” can also order website hosts to take down unwanted content, says the manager, who trains officers on the company’s products. Typically, the officers will run their keywords on search engines such as Google or Baidu to see which websites or postings get the top search rankings. “If there is a subversive comment [the officer] will tell the web host to block it or to erase this posting.”

Elsewhere, government departments monitor the online response to their policies and watch out for unrest brewing in their area of responsibility, or for accusations of misconduct or corruption against one of their own. This information is then – selectively – passed on to the local propaganda department and information offices, which decide on a response. This might include dictating to state media the line to be taken on the issues at hand, or instructions to websites about which news items they may run. More often than blocking a news item entirely, the departments will instruct websites to keep coverage short, and bury it.

With this workload, the government would have to have millions of web censors. Yet Gui Boachen, head of the internet department at the Communist party branch in the city of Luoyang in Henan, China’s most populous province, says his office and the information office, the two principal internet surveillance authorities in the city, have only 20 people watching the web – a tiny number for a city where 300,000 websites are registered.

. . .

When I flew from Beijing to Kunming, capital of Yunnan province, to see Wu Hao, two of his aides were waiting at the airport to take me to a five-star hotel on the outskirts of the city. My room – reserved by Wu’s department: I wouldn’t be invited to the office to poke around – was on the top floor. Its bay window overlooked upscale condominiums, a blue lake and towering mountain ranges on the horizon. Wu came to meet me there.

His motivation had little to do with explaining how the system works; that information came out in passing. Rather, he wanted to impress on me the fact that the Chinese government is moving towards transparency and pushing local officials to embrace these values as well. The government was fighting, Wu said, for a free, critical media. But when Wu described his blueprint for reform, this hardly seemed the likely result.

The current online content reporting system allowed individual government departments to cover up problems, he complained. Instead of that system, he wants to establish a public opinion “situation centre” which would enable the provincial government to come up with emergency responses to criticism spotted online before the matter gets out of hand – a little like what happened in the case of Li Qiaoming and the “elude the cat” scandal.

The internet is “an instrument of raising our governing ability”, he said. If necessary, the censors will “guide” the media for this end. An example of this was seen in May, in a case that had little to do with the internet. Wu called in local media to discuss a court case against companies accused of having polluted Yangzonghai, a large lake in Yunnan. The media had allowed the companies space to present their case. This “created a bad influence on the judiciary”, says Wu. “The people also got confused whether the lake was actually polluted by [the defendants],” and the government started worrying that unrest would spread among the 20,000 residents along the lake who were hoping for compensation following a guilty verdict. Therefore, the authorities came to an “agreement” with the media on more “balanced” reporting of the case.

Wu claims that such interference is just a stage in the transition from traditional Leninist propaganda work, where the media are part of the government and party apparatus, to a modern, transparent system where the government communicates policies to an independent media through one institution, and regulation is done through another, independent body.

It might have seemed like a reformist’s message – until he explained which institutions would take on which responsibilities. The government’s information offices would provide public relations and spokesmen, while the Communist party’s propaganda department would take over the role of independent media regulator.

. . .

China’s 350 million internet users are a fast-moving crowd. They increasingly create their own online content and sign up to social networking sites – two of the tenets of the so-called Web 2.0 revolution. They use YouTube, Twitter and Facebook – or their Chinese equivalents – instead of old-style e-mail and search engines. There were as many as 162 million blog authors in China by the end of 2008, and the number of community websites and bulletin boards is growing exponentially.

These changes have led to a new style of censorship – outsourcing, or what is commonly called “Censorship 2.0”, in which bloggers paid by the government aim to neutralise debates that the authorities don’t like. In some ways, this isn’t new: every internet portal already self-censors. (There are no set rules for this. A posting removed from one website might survive on another, and blogs closed by one host regularly move to competing ones.) Now, however, “50-cent bloggers” – named after the price paid per posting when these freelancers first appeared – sign up to chatrooms or bulletin boards and speak up for the government, or against its critics.

Two months ago, for example, on Tianya, a popular bulletin board website, a conversation developed about the death of a young man who had been hit and killed by a speeding driver with a record of illegal car-racing. The debate focused on the fact that the police failed to challenge the driver’s claim that he had been travelling at 70kph – at most. Seven participants then started attacking those who were challenging the police. It could be seen from their sign-up data that all seven had registered at Tianya within the same two days.

In Beijing, internet surveillance authorities provided even clearer evidence of the use of paid bloggers by publishing a call for volunteers to sign up as “internet debaters”. Applicants had to answer a series of questions regarding their personal background and political views.

Wu admits that it is part of his responsibility to “balance” online opinion. “The internet is a platform where anyone can express their opinion. When there is the situation that opinion leans totally to one side, then we will indeed put some different voices out there to allow the public to make their own judgment independently.”

It is this practice of dragging ordinary citizens into the censors’ embrace that makes China’s internet surveillance system so sinister – and what makes it so hard to predict whether the government or the voices of dissent might prevail in the end. Analysts call this surveillance strategy a ­“panopticon”, in reference to the 18th-century English philosopher Jeremy Bentham, who thought up a prison design which would allow monitoring of all prisoners at all times without allowing the prisoners to know whether they were being watched or not. When internet users are involved in anonymously watching each other, this becomes a participatory panopticon.

“Maybe you imagine things as a battle between good and evil, between the good netizens and the evil censors. It’s not that simple,” says the editor of a Chinese magazine focusing on culture and society. “As a Chinese, you are always automatically part of ‘them’.”

. . .

Throughout my conversation with Wu, one fear became clear: that information, released in unprecedented quantities and at unprecedented speed through the internet, could eventually end the Communist party’s monopoly on power. “The herd instinct on the internet is very severe,” he said. “An opinion, put online, can create a following, a magnifying effect.”

Will the people that the party rules one day respond to a different rallying cry, passed on through the web? It’s not unthinkable. Over the past two years, China has seen a surge of cases where an obscure incident, made public over the internet, has led to popular dissent. In October 2007, forestry authorities in Shaanxi province caused outrage by handing a Rmb20,000 (£1,800) reward to a local farmer who claimed he had sighted a rare tiger in the woods but whose photographs of the beast were exposed as fakes. It was said that the officials involved, 13 of whom were sacked, were hoping to attract investment and tourism to the area. In June last year, the police’s unconvincing explanation for a teenage student’s death in Guizhou led to widespread questioning and ridicule of the local administration, culminating in the torching of government buildings and police cars in Wengan township. And when, a month later, Yang Jia, an unemployed man who said he had been mistreated by police after using an unlicensed bicycle, stormed into a Shanghai police station and killed six officers, countless netizens hailed him as a hero.

The examples continue: in November last year, Lin Jiaxiang, a Shenzhen official, was caught harassing a young girl and then berating her father when he complained; Lin became the target of fierce criticism in blogs and chat rooms and was demoted. In December pictures posted online of Zhou Jiugeng, a housing official in Nanjing, wearing what was called a luxury watch and smoking expensive cigarettes, led to his demotion.

These cases, says Wu, show that the authorities need to build an ever-more sophisticated system of interacting with those who form public opinion on the internet. “I know a group of people in Yunnan who live for the internet, have no respectable job but just express their views online on everything and anything,” he says. Despite the contempt in his words, he claims that he can co-opt such figures by means of modern public relations – setting the agenda, leading the debate.

. . .

Government critics question Wu’s confidence in his strategy. “The cost of doing that is rising because netizens are getting ever more experienced in fighting the censors,” says Zhou Shuguang, a young man in Hunnan who blogs under the name Zola.

In recent months, the government itself has shown it shares those doubts. Since the March anniversary of the unrest in Tibet, the censors have fallen back on blanket blocking of certain websites such as foreign media outlets and YouTube – crude measures that Beijing was meant to have left behind. As the government prepared for the 20th anniversary of the bloody suppression of the Tiananmen democracy movement on June 4 1989, more shutters went down: Hotmail, Twitter and other applications used by tens of millions of Chinese became unavailable for a week. And when a wave of ethnic violence swept China’s westernmost region of Xinjiang two weeks ago, the government even went as far as taking the entire province off the internet and partly shutting down mobile phone services.

The effort to force computer manufacturers to pre-install content-control software on PCs sold in China makes a mockery of government claims that it was moving towards a more open, free and transparent media environment. In fact, the government’s apparent climb down on the matter suggests it may have to rely more on 50-cent bloggers and other Censorship 2.0 techniques than the blunt force of technology.

An executive working at one of China’s leading internet portals tells me: “The task [for the Communist party] has been to allow enough noise in the system for people to let off steam and make them feel that they are living kind of a free life, but at the same time to maintain a sense of fear and respect that keeps them from demanding big change.”

Meanwhile, internet use is undergoing another profound shift as it extends beyond its predominantly urban base. Already, by the end of 2008, 117 million people, or more than one-third of the country’s web users, could access the internet on their mobile phones. A rapidly growing number use QQ, China’s largest online messaging tool, on their handsets wherever they go. That means that the vast rural hinterland, where about 70 per cent of the population still live, is getting a fast-track link-up to a network via which they can voice their dissatisfaction: about corrupt and despotic officials, unsolved problems of pollution and social security, land grabs and income disparity.

“That is the perfect design for jump-starting social unrest in rural China,” says the internet portal executive. “And I think the chain of ­‘internet mass incidents’ over the past two years is an indication that it has started to work.” The government may have previously been effective in curbing online chatter, he says, “but they didn’t expect the repercussions of all that online noise in the real world”.

Kathrin Hille is an FT correspondent in Beijing