How China gets the Internet to censor itself

    In 2006, these two animated characters, Jingjing and Chacha, appeared on websites in Shenzhen, China, to remind Internet users that they were being monitored.

    Just who owns the Internet, and who has the right to control what content is available on it? Is it sovereign territory, or is it free from the confines of antiquated earthbound laws? These questions have engaged Internet activists and scholars for over a decade. And, after the intense debate last month over proposed Internet restrictions in the U.S., announcements from Twitter and Google about enhanced efforts to voluntarily comply with national laws, and ongoing international protest over the Anti-Counterfeiting Trade Agreement, popular interest in Internet regulation appears to be mounting.

    To the disappointment of techno-utopians, though, the Internet is very much capable of being regulated, and some governments have been perfectly willing to do so. The most obvious example of this is China’s “Great Firewall,” a vast network of structural, social and legal controls by which it regulates Internet content. Exactly what content is being blocked, however, isn’t always easy to say.

    One can intuitively guess that advocating the overthrow of the Communist Party online would be difficult, if not dangerous, to do in China. But is Taiwan a sensitive topic? What about religion and sex? For several months last year, I set out to track what one Chinese Internet company, Sina Weibo—China’s leading Twitter copycat—considered off-limits. Utilizing a computer script and much patience, I was able to uncover roughly a thousand unique banned words. According to that list, I can tell you that Taiwan is mostly fine so long as you’re not discussing Taiwanese independence (台湾独立, 一中一台, etc); all discussion of major religions is allowed except for one, Islam (伊斯兰); and even today, in an age of increasingly open sexuality in China, searching for posts on aphrodisiacs (春药) will return error messages.

    Chinese Internet Regulations

    Weibo is a general term for microblogging—literally, “tiny blog”—representing a whole host of Twitter-like clones in China. However, Weibo has become synonymous with the largest microblogging site, Sina Weibo. It wasn’t the first, but it is by far the largest and most important such site in China. Aided by China’s banning of Twitter, the site has grown to over 250 million registered users.

    However, like all major licensed websites in China, Weibo has numerous restrictions on what sort of content it is allowed to host and distribute. In June 2010, China’s State Council Information Office released a white paper on Internet usage for the country. Though the paper asserts that Chinese users have the right to freedom of expression online, it also enumerates a prohibition against content that is:

    endangering state security, divulging state secrets, subverting state power and jeopardizing national unification; damaging state honor and interests; instigating ethnic hatred or discrimination and jeopardizing ethnic unity; jeopardizing state religious policy, propagating heretical or superstitious ideas; spreading rumors, disrupting social order and stability; disseminating obscenity, pornography, gambling, violence, brutality and terror or abetting crime; humiliating or slandering others, trespassing on the lawful rights and interests of others; and other contents forbidden by laws and administrative regulations.

    This is an incredibly broad array of off-limit topics, and the fact that a phrase like “damaging state honor and interests” is not clearly defined is an intentional feature of the Chinese censorship system, a mechanism coined by Perry Link as “the anaconda in the chandelier.” The vagueness inevitably leads content providers like Sina Weibo to excessively self-censor in order to stay well within the bounds of acceptable discourse. The company—and its users—may have a sort of sixth sense for knowing what may or may not be off-limits, but the fact that there is no officially published blacklist, coupled with the fear of severe punishments, compels them to step even farther back from the imaginary line. As Internet scholar Rebecca MacKinnon noted:

    Recent academic research on global Internet censorship has found that in countries where heavy legal liability is imposed on companies, employees tasked with day-to-day censorship jobs have a strong incentive to play it safe and over-censor—even in the case of content whose legality might stand a good chance of holding up in a court of law. Why invite legal hassle when you can just hit “delete”?

    Chinese Internet companies are now required to sign the “Public Pledge on Self-Discipline for China Internet Industry,” a document that in some ways includes even stricter rules than those listed in the 2010 white paper. Thus, it’s no wonder you have companies censoring topics like Islam, even though the religion is officially sanctioned under Chinese law.

    Chinese government officials send weekly updates to media providers on topics that it expects censored. Otherwise, however, the onus is on the content provider to self-censor, a practice that Weibo’s head editor admitted is “a very big headache.” But it works. Bill Clinton may have compared censoring the Internet to nailing jello to the wall, but China appears to have built an effective harness (self-censorship by companies and netizens) to go along with the world’s biggest nail gun (tens of thousands of state-employed Internet monitors, total government control of overseas Internet data connections, and next-generation monitoring hardware developed by corporations like Cisco).

    How One Company Censors

    When I say “banned” with regards to words on Weibo, I actually mean “blocked.” Users can actually post anything they want to the site, including words which are indeed off-limits. Some other sites, in contrast, employ filters that will deny you the ability to post a message if you use a banned word. Weibo’s censors can then summarily delete inflammatory messages without any notice. Unless the author is very notable or the post is caught up in a roundup, it is likely that posts with banned words might escape the censor’s eye.

    For the most part, however, Weibo relies on a different strategy for containing the spread of sensitive content: blocking searches. By blocking a user’s ability to find a term, one isn’t able to look for sensitive content and one doesn’t have to delete or filter posts one at a time. Not only is this method arguably more effective and more flexible, it’s less intrusive. Users facing an error message when posting their own content might feel outraged, but being unable to search for a term would probably just elicit a shrug. And words that are only temporarily sensitive can be added to the blacklist of search terms one day and removed the next without having to permanently delete the content.

    When censors decide a certain search term is no longer sensitive, as they did for hundreds of words like 恋足 (foot fetish) and 九一一袭击 (the 9/11 attacks) in late January, the switch is flipped and users can suddenly search for foot fetish posts to their heart’s content—so long as they haven’t been intimidated by the chilling effects of the previous block. I haven’t used the service enough to know how Weibo handles “re-tweets” and sensitive posts from flowing into a follower’s feed, but there are reports that Weibo sometimes will simply “ghost” a post—allowing it to appear online to the author, but not be viewable to anyone else—rather than delete it. Such a tactic demonstrates how concerned Weibo is with maintaining a seamless user experience while also having to comply with content restrictions.

    At the moment, Weibo’s search-filtering mechanism is not particularly sophisticated. It checks the search term against a blacklist, and if any part of the search term matches any word on the blacklist, it is blocked.  For example, “Nintendo 64” is blocked because “64,” short for June 4, the day of the 1989 Tiananmen crackdown, is a banned term. Thus, any search containing “64” will be blocked, even harmless ones like “Nintendo 64,” an issue known as the Scunthorpe Problem.

    Over the years, in a series of cat-and-mouse games, Chinese Internet users have developed an extensive series of puns (both visual puns and homophones), slang, acronyms and memes to skirt restrictions and censors. Such creative usages may still be helpful in evading the censor’s eye on Weibo—by using a sort of code it makes one’s post not only less likely to get caught in any search filter but also less likely to even be found by a censor in the first place. Furthermore, Chinese Internet users have seemingly mastered the use of irony as protest, reaching the point where any decidedly pro-government comment online like “Socialism is good” or “I have been represented by my local official” must be assumed to be satirical. Filtering tools like the ones Weibo uses in its search tool certainly can’t recognize such nuances. In some respects, they are “easy” to defeat, thus emphasizing just how important those human monitors employed by Weibo are, who have the ability to delete individual posts and even entire accounts, like what happened to Ai Wei Wei’s.

    Blocked on Weibo

    Inspired by China Digital Times’ impressive attempts to track banned words across various Chinese online services, I sought to systematically uncover banned words on Weibo. I used 700,000 Chinese Wikipedia titles as my search terms (which include both traditional and simplified titles as well as many colloquial terms) and tested them on Weibo with a computer script. In the end, I came up with several thousand blocked terms, roughly a thousand of which were unique. Of course, some were unsurprising, for instance political terms like 六四 (64) and 反共 (anti-communism); others like 乱伦 (incest), 暴露狂 (exhibitionism), and 吹箫 (blowjob, but literally blowing a flute) straddled a moral gray area; a few like 伊斯兰 (Islam) and 同性爱 (homosexuality) were surpising in their reactionary nature; and finally, a few words like 黄色 (yellow, slang for pornographic) bordered on ridiculous (mercifully, yellow has been unblocked since early February).

    In the small sample that I’ve coded so far, the largest share of the blocked words are names of people, the majority of which are Communist Party members—protection from criticism on Weibo being a seeming perk for rising up the ranks—while dissidents and people caught up in scandals or crimes made up the rest of the blocked names. Words related to sex, immorality, religious cults, demonstrations and protests, also comprised a large percentage of the list.

    I was aided in all this by the fact that Weibo notes when it has in fact blocked your search. In addition to returning zero results, it helpfully displays an error message. Thus, one is literally aware when search results are blocked, unlike instances on other sites when China’s firewall may leave a user ignorant that his connection and searches are being filtered or degraded (though a handful of searches on Weibo—for example 无界网络, or, as it’s known in English, Ultrasurf, the Falun Gong-designed anti-censorship software—do cause your connection to the site to be intentionally cut for two minutes).

    Such “transparency”—a notice posted when content is blocked—is the same kind promoted by Twitter as a check against censorship. And while transparency is generally laudable, it could be argued that a visual reminder that censorship is occurring serves as a form of intimidation, akin to the appearance of the animated police characters Jingjing and Chacha on various Chinese websites in recent years. Such transparency also serves as an effective training mechanism, teaching Internet users about which topics they should be careful of discussing online, thus furthering the goal of decentralizing the censorship and moving the onus for it from the government to the media company to, finally, the individual.

    Thus, there are multiple layers of censorship occurring. There is the government mandated blacklist of off-limit topics—what we’d typically consider censorship—but there are two more subtle forms: the self-censorship by content providers, who must make judgment calls on what needs to be censored in order to stay in the government’s good graces, and self-censorship by users, who face the threat of being detained and punished for anti-government posts. Users are at greater risk than ever now that Weibo and other micro-blogs are beginning to require real names during registration. Though the company and government claim that this is merely to hold users accountable for spreading misinformation and malicious rumors, it’s clear that such a measure is designed to head off the type of political commentary that could lead to an online-inspired Jasmine Revolution. Even so, Internet users are clever, and with ever-growing information about how companies and governments censor content online, the mice may never be caught.

    Recent Stories

    • Feature

    How Russians, Indigenous people and Belarusians are uniting to resist the war in Ukraine

    January 26, 2023

    Antiwar activists in Russia are finding support and solidarity in a growing resistance network comprised of Russian diaspora, Indigenous and ethnic minorities and Belarusians.

    • Analysis

    How New Yorkers can remove George Santos and root out Trumpism

    January 23, 2023

    New York may have cost Democrats the House, but grassroots groups across the state have the antidote to a politics of racial division and control.

    • Breaking News

    Movement to stop Atlanta’s ‘Cop City’ calls for support after police kill forest defender

    January 20, 2023

    Atlanta activists are calling for an independent investigation and solidarity, after police killed an Indigenous land defender in a heavily-armed raid.