Slashdot Mirror
China Is Censoring People's Chats Without Them Even Knowing About It (qz.com)
An anonymous reader quotes a report from Quartz: A new study from The Citizen Lab, a research group at the University of Toronto, reveals that censorship on WeChat occurs primarily in group chats rather than one-on-one chats between two people, and often in such a way where the sender of a text isn't even aware a piece of text has been scrubbed. The discoveries illuminates how China's government attempts to keep its citizens blind to the scope of its censorship regime. The researchers set out find the extent to which certain keywords got scrubbed from conversations between two or more users in WeChat. To do this, in June 2016 the team posed as a Chinese WeChat user and sent out 26,821 keywords containing terms that had been censored on other apps, including Tom-Skype (a made-for-China version of Skype) and YY (a live broadcast app). A corresponding Canadian user in the two-way chat would then report back to say whether or not the message had been received. The report states that out of the entire sample, only one term -- Falun Gong -- had been scrubbed. When they ran an identical test in August, even that text mysteriously passed without censorship. Yet when they tested group chats, they found multiple cases in which certain keywords triggered a removal. Specifically, while sensitive terms used in isolation were unlikely to trigger censorship (say "June 4th," a reference to the Tiananmen Square protests, brutally put down on June 4, 1989), it took effect when they were used in a full sentence or with other keywords. The researchers also discovered that when WeChat censored a message, the sender received no notice informing him that his text had not reached the intended recipient. The study also notes that "WeChat only censors content for users who bind their account to a mainland Chinese phone number when they first register to use the app." The censorship is still applied even if Chinese residents move to different countries or change phone numbers.
The words "orange" "tiny" and "bankruptcy" will all similarly be removed without any notice.
1: Censors remove all references to June 4th, no matter what context
2: Any plans for business to take place on June 4th disappear
3: Eventually nothing gets done on June 4th because nobody can plan for it
4: It turns into a holiday, and nobody can remember why.
Some 15 years ago I ran a messageboard for a popular game. It suffered from some persistant trolls and spammers. When they were banned they would come back with a new account, all the time from different IP's. After some time I made it so that instead of blocking access for the banned accounts or IP's, I hid their posts for all other users. Some clued in, but it did work against most. Later I made it so that the moderators would still see the banned posts and I asked them to "feed" the trolls, to make it harder for them to realise something was wrong.
It was quite a lot of work to adapt the code, but it was an interesting experiment. I always thought it would be a quite useful tool for bigger sites, so I'm surprised I never saw it used anywhere else.