Chinese National Firewall Isn't All that Effective
The study, carried out by graduate student Earl Barr and colleagues in the computer science department of UC Davis and the University of New Mexico, exploited the workings of the Chinese firewall to investigate its effectiveness.
Unlike many other nations Chinese authorities do not simply block webpages that discuss banned subjects such as the Tiananmen Square massacre.
Instead the technology deployed by the Chinese government scans data flowing across its section of the net for banned words or web addresses.
When the filtering system spots a banned term it sends instructions to the source server and destination PC to stop the flow of data.
Mr Barr and colleagues manipulated this to see how far inside China’s net, messages containing banned terms could reach before the shut down instructions were sent.
The team used words taken from the Chinese version of Wikipedia to load the data streams then despatched into China’s network. If a data stream was stopped a technique known as “latent semantic analysis” was used to find related words to see if they too were blocked.
The researchers found that the blocking did not happen at the edge of China’s network but often was done when the packets of loaded data had penetrated deep inside.
Blocked were terms related to the Falun Gong movement, Tiananmen Square protest groups, Nazi Germany and democracy.
On about 28% of the paths into China’s net tested by the researchers, blocking failed altogether suggesting that web users would browse unencumbered at least some of the time.
Filtering and blocking was “particularly erratic” when lots of China’s web users were online, said the researchers.