The Great Firewall of China isn't impenetrable—it just takes a little elbow grease and high Internet traffic to squeeze a few banned terms through. Researchers at the University of California at Davis and the University of New Mexico have made new findings on the inner workings of China's firewall, designed to keep citizens "safe" from words, concepts, and events that the Chinese government does not consider to be good for its citizens. What they found showed that the firewall wasn't as sophisticated as it's cracked up to be, but it did a good job at working some social engineering magic to scare citizens into avoiding banned terms.
The group based its methods for testing the words on a 2006 University of Cambridge finding that the Chinese firewall sends a series of resets to both the source and destination when it detects a banned word, which ultimately returns a broken web page to the user. This finding allowed the UC Davis and UNM researchers to test which words were blocked by sending out carefully chosen words and seeing which ones were being sent back. The researchers chose words from the Chinese version of Wikipedia and, using latent semantic analysis in their tool called ConceptDoppler, found related words to test as well. Some of the blocked terms included information on the Falun Gong, Nazi Germany "and other historical events," and "general concepts related to democracy and political protest."
The team, led by UC Davis Computer Science grad student Earl Barr, found that China's firewall did not actually stop banned terms at the front lines all the time—in fact, many banned terms made it through several layers of routers before being "returned to sender." The firewall also accidentally let banned terms through to the end user about 28 percent of the time, which the researchers said was particularly erratic during high-traffic times.
Barr theorizes that the occasional slip encourages China's citizens to engage in self-censorship, however, like a panopticon. If they see terms that are blocked most of the time, it might remind them that their traffic is being monitored and filtered regularly. In theory, this would pressure Chinese citizens into avoiding those terms during their normal Internet travels for fear of being watched by the government, he said.
That interpretation may not be too far off, as the Beijing Municipal Public Security Bureau recently announced that it would begin putting animated police characters on all government-approved web sites in order to remind citizens to behave in cyberland.
The team plans to present its findings at the ACM Computer and Communications Security Conference in late October.
reader comments