Issue #11 Creative Ways to Dodge a Chinese Internet Crackdown
This is issue #11 of the Magpie Digest newsletter, originally sent on 5/17/2018
Like the tides or the trendiness of bell bottom jeans, the Chinese government’s efforts to control culture and speech ebb and flow on their own schedule. Given the current political climate (domestic regime stabilization + international tension from multiple sides), it is no surprise that 2018 has already been a busy year for the censors.
But each crackdown is just the start of a cat-and-mouse game between regulators, platforms, and a very resourceful user base. This well-tread exchange is as old as online censorship itself, with users discovering clever ways to circumvent new censorship technologies as they emerge in real time. This week, we take a look at some of the common circumvention practices that have emerged.
Understanding Online Crackdowns
We’ve written about the capricious nature of top-down, government-imposed cultural bans before, but let’s lay some groundwork on crackdowns — a slightly different beast — before we dig in.
We are using the term social media crackdowns to describe real-time responses to specific online conversations deemed to be unwieldy (as in the case of #MeToo and the kindergarten abuse scandal we covered). These are usually implemented directly by social media platforms at the behest of the government.
The goal of a crackdown is to disperse the digital crowds gathering around a topic, and to make it difficult for that conversation to attract more attention. Tactically, this means takedowns of core posts that the conversations are rooted around, filtering keywords, hashtags, and specific pieces of media from being posted or searched at a technical level, and suspending, banning, or even “drinking tea” (a threatening visit from a government official) with highly visible users. As the objective is to prevent large-scale social mobilization, the system is most frequently deployed against news and conversations that are likely to incite mass outrage, especially against government officials, regulations, or anchor institutions.
It’s also possible for crackdowns to be the result of overaggressive self-censoring by platforms — for example, when Weibo faced major backlash from its users last month for including gay content in a crackdown on violent and pornographic content, government mouthpieces like the Communist Youth League account and the People’s Daily newspaper were quick to publicly disagree. Weibo ended up walking back the decision.
With the exception of some topics that have been deemed permanently sensitive, this form of censorship is more like bad weather than it is a permanent wall — after the moment passes and the conversation dies down, the blocked keywords and sometimes even banned accounts are often restored.
Circumvention Tactics Co-Evolve Alongside Censorship Tools
The oldest and simplest form of crackdowns is a blacklist of temporarily sensitive keywords — hashtags, catchphrases, names, and other search terms around a topic — that flag existing posts for removal, cannot be included in new posts, and are wiped from search results and trending topics.
To buy a conversation more time once the crackdown has started, users circumvent this text-based filtering in two ways:
1) By using homophonic puns — a linguistic resource in which written Chinese is unusually rich — and other wordplay to create easily-recognizable substitutes for blacklisted keywords. A recent, highly publicized example was the widespread use of the Weibo hashtag #米兔# (mi tu, literally “rice rabbit” but also a homophone for #MeToo) to circumvent the crackdown on discussion around sexual assault on university campuses;
2) By posting screenshots of the offending text instead, since images are harder to automatically filter than text.
As image and text recognition capabilities have gotten more sophisticated over time, platforms now include specific images in crackdowns. But armed with a basic understanding of how that technology works, users are exploring new methods of circumventing that as well. After Peking University student Yue Xin’s open letter calling for an investigation into an old case of sexual assault at the university after was taken down from WeChat and Weibo, users found that they could not upload screenshots of her post to these platforms. Undaunted, many uploaded distorted and remixed versions of the screenshot to successfully thwart the image and character recognition algorithms.
In the aftermath of Yue Xin’s letter, one user figured out a way to post its content in a way that is technically impervious to takedowns altogether: as the note accompanying an Ethereum transaction. Though China is moving quickly to regulate cryptocurrency, the nationwide obsession with the blockchain as both a technology and financial opportunity ensures that plenty of people will have ready access to the letter forever as long as they know where to look for it.
As on the Western internet, the savvy to navigate platform-wide crackdowns is unevenly distributed. In China, the unlikely group that is among the most advanced and practiced are fanfiction writers and readers.
Since its launch, Weibo has been a hub of fandom activity in China — as well as the site of the cat-and-mouse game between the platform’s strict rules on sexual content and racy fanfic writers. As we described above, posting text as images was an early strategy to take root. Rather than just posting a plain screenshot, writers began to optimize for these new constraints, creating beautifully laid out and typeset pages designed to look great on a phone. This became such a genre convention that even safe-for-work fanfiction has adopted the format.
As Weibo has gotten more sophisticated at blocking these posts, possibly through character recognition, the community reacted by moving their work to other platforms and simply posting the links on Weibo. As each platform gains enough momentum to warrant the scrutiny of the censors, they migrate again. At time of writing, a common practice is for fanfiction writers to post their stories as pictures in unlisted documents on Shimo, a collaborative editing tool originally designed for professional use, and then to share the links to Weibo. Weibo has nothing to censor, and Shimo does not have an easy way of knowing that violating content on their site is being read by thousands.
For China’s savvy internet users, these feats of nomadic platform acrobatics have become a way of life.
As QZ reports here, another tactic Chinese internet users are employing to bypass text-based filters is an old form of internet writing called 火星文 (huo xing wen, literally “Martian language”) — think a very advanced form of Chinese 1337speak that uses numbers and obscure characters in place of standard characters. If you can read or write Chinese, give this Martian translator a spin to behold the horrific ways you can put Chinese characters through the wringer.
If you enjoyed this Magpie Digest, please sign up to receive it in your inbox every week or pass it along to a friend!
Magpie Digest is a project of Magpie Kingdom, a consultancy that provides analysis, business advisory services, and custom research to help businesses translate their value for the Chinese market. The Magpie team (Christina Xu, Tricia Wang, and Pheona Chen) is based in New York and Shanghai.
Hire Us: The Magpie team is available for strategic advising, speaking engagements, and custom research projects. Email us with inquiries.