Chinese Blockage: That Country Needs an Enema

The struggle for control of the Chinese Internet continues. We reported last week that two dissidents, in custody for “subversive” Internet use, died in custody under mysterious circumstances. Last month we discussed the “hacktivists” seeking to liberate users from blockage. Now, Jonathan Zittrain and Benjamin Edelman of the Berkman Center for Internet & Society at Harvard Law School have issued a report called Empirical Analysis of Internet Filtering in China, which concludes:

From our data, it appears that the set of sites blocked in China is by no means static: whoever maintains the lists is actively updating them, and certain general-interest high-profile sites whose content changes frequently appear to be blocked and unblocked as those changes are evaluated. (This is particularly noticeable with news sites such as CNN and Slashdot.) Some new sites with sensitive content do not appear to take long to be blocked. However, even some longstanding sites of apparent sensitivity remain unblocked. This is most easily noticed in our data with respect to sexually-explicit sites — we found blocking of only 13.4% of our sample of well-known sexually-explicit sites — but is also anecdotally apparent from our data, as one notes blocking of some US intelligence sites but not others, etc. Further data collection will be geared at determining the extent to which the basket of sites blocked reflects shifting substantive government policies — whether, for example, a sea change in relations with Taiwan, whether positive or negative, is reflected in blocking, and if so, how quickly.

China’s Internet filtering efforts remain opaque, and in the absence of government cooperation or admission of filtering methods, data probing of the sort used in our study remains a useful tool in determining the scope of filtering. The authors have previously studied filtering in Saudi Arabia and in American public libraries; in these locations, blockage of a web page leads to an error message clearly explaining that the requested page is unavailable due to intentional blockage. In contrast, China’s systems make it difficult for a user to distinguish between an intentional block and a temporary network or server glitch. This may be intentional or may reflect technical happenstance — that this implementation was easier or cheaper, given the size and design of China’s network infrastructure. But some newer forms of Chinese filtering — namely, redirection of a request for a sensitive web site to another web site — can be either more or less obvious to the user than an apparent network glitch, depending on whether the substitution is noticed.

The primary and most longstanding means of blocking is at the router level, and on the basis of IP address — the crudity of which means that those implementing filtering must choose between blocking an entire site on the basis of a small portion of its content, or tolerating such content. This would explain why, for example, the www.mit.edu server is sometimes wholly inaccessible even though Chinese officials likely have no objection to most content on that server. To the extent that the entirety of that server is nonetheless inaccessible, China’s filtering system is properly considered to be overblocking, and we believe our data indicates extensive overblocking of this form. This may account for the rise of still-rare forms of blocking that allow more refined content filtering — such as blocking by keywords or phrases in any particular HTML page requested by a user, whether or not the site hosting the page is present on an ex ante block list. Such blocking is likely far more technology-intensive, in principle even slowing overall network response time as packets are analyzed by sniffers and the results passed to filters. Aside from allowing more refined content filtering, such newer forms of blocking appear to be linked to disabling Internet access for an arbitrary amount of time for a user who requested a page with forbidden content — enabling a penalty for attempting access to sensitive material beyond simply denying the very material requested. Other nascent but growing forms of filtering appear to be targeted to limit the information that can be gleaned from search engines — enabling the automated blocking of search results that may not (yet) have been filtered through human placement on a “forbidden” list.

The Chinese government and associated network authorities are clearly continuing to experiment with different forms of blocking, indicating that — unlike Saudi Arabia, which appears to have a single, declared method of blocking and a much more constant (and apparently smaller) list of non-sexually-explicit blocked sites — Chinese network filtering is an important instrument of state Internet policy, and one to which significant technical and human resources continue to be devoted.

Obviously, on the political front, the Chinese government stills feels it has much to lose from the free flow of information to its people