Bypassing Chinese Censor

It is well known that the Chinese government proactively clamp down on any media that it deems to run contrary to its own interest. Besides keeping a tight lid on official press such as newspapers and TV reports, popular online forums are also actively patrolled to keep a tight patrol on dissidents that run contrary to their appetites.

Chinese netizens, of course, aren’t simply willing to lie low and submit to that. Apart from technical solutions such as proxy surfing, they’ve also been quite creative in forum posts. Chinese forums are commonly littered with different-but-phonetically-similar words in place of more sensitive terms. But that too, gets caught by the officials pretty quickly.

The latest round in this cat-and-mice game is rather ingenious – rearranging words to read from a different orientation to fool the (very likely computerized) scanning. For instance, instead of writing from left to right, the posts are authored from right to left, or even scrambled to a top-bottom arrangement. This, in fact, heralds back to the traditional Chinese script back when the times where it was written on bamboo strips vertically from right-to-left. Thus, it makes perfect sense for human readers, but State computers may have a tougher time combing.

An example of the scrambling software:


Which made me ponder – captchas have come to become an integral part of the Internet landscape to verify if one is a human or not. This could mean having to recognize jumbled-up digits or alphabets, or recognizing cats from dogs. These are tasks deemed easy for the average human being, but tough for computers to crack.

Could there be other similar hacks around censored forums that are easy for humans to decipher, but difficult for computers to crack? What are some of the methods that defies simple computing power? And tools to achieve this in a forum setting? Scrambling orientation is one of them – could there be more (and more interesting ones too)?

[via WSJ]

Bookmark the permalink.

Leave a Reply

Your email address will not be published.