John Palfrey

The OpenNet Initiative’s Methodology for Studying Internet Filtering in China.

Members of our consortium have been collecting data on China’s Internet
filtering regime since 2002.  The data included in this report have
been updated as recently as this week.  As the Chinese government has
developed more sophisticated means of filtering, we too have developed
more sophisticated and comprehensive means of testing their filtering
efforts.  Since our last study, our testing methods have become
substantially more fine-grained and reliable. 

To gauge how Internet filtering likely affects the average Chinese
Internet user, ONI employs a variety of means to test blocking and
censorship and to ensure data integrity.  We test filtering from
different points on China’s network, in different geographic regions,
across time.  The resulting data allow us to conduct rigorous
longitudinal analysis of Internet blocking in China.  We examine both
the response that users receive from the network and from the Web
servers involved and information about the route that a request takes
on its way from a user to a Web server – allowing us to pinpoint
exactly where information is censored and controlled.  While it is
impossible to paint a flawless picture of China’s Internet filtering
efforts at any given time, we are increasingly confident that our data
present an accurate snapshot of China’s Internet filtering regime today.

We have tested China’s Internet filtering regime using four methods. 
Under Nart Villeneuve’s leadership, ONI developed and deployed an
application to test within China what content is, and is not, blocked
by the state’s system.  Volunteers installed and ran this application
on their home computers to allow ONI to probe China’s filtering from a
wide range of access points inside the country.  Our volunteers also
ran manual checks for access to web sites.

Second, we accessed proxy servers in China to duplicate and augment
this in-state testing of whether or not a citizen could access a
certain web site.  Proxy servers are points in China’s network that act
to aggregate and respond to user requests for content.  Accessing a
proxy server in China allows ONI to browse the Internet as though we
were in China, even though we are physically located in another
country.  Through proxies, we are able to obtain a random sampling of
Web content – and censorship – across multiple networks and service
providers.

We have also explored whether China blocks other types of
Internet-related communications.  Anecdotal evidence has suggested for
a long time that China blocks certain e-mail communications and that
Web logs – or “blogs”, which are personal online journals, often kept
by increasingly famous activists – have been more recently targeted by
the Chinese government for blocking.

To test these hypotheses, we published content on blogs on three of
China’s most popular blog providers to evaluate the services’ keyword
filtering mechanisms.  We then later sought to access this blog content
that we had published.  

Finally, we sent a series of test e-mail messages to, and from,
accounts hosted by several Chinese ISPs.  These messages contained
content on sensitive topics – such as political dissidents, objections
to the state’s repression of the Tiananmen Square protests, and
religious persecution – typical of e-mails sent by human rights
organizations.

In addition to employing these technical methodologies, we have closely
studied the legal and policy regimes in place in China.  The insights
of many scholars and activists, both inside China and elsewhere, guided
our research and provided quality assurance.

Topics Censored by the Chinese Filtering Regime.

China filters Internet content on a broad array of topics.  The censors
particularly target sensitive political topics for blocking.  To
determine precisely what is blocked, we created a keyword list of terms
on sensitive topics, such as the Falun Gong spiritual movement, the
Taiwanese independence movement, and criticism of China’s government
and leaders.  We used the Google search engine to compile a list of
large numbers of sites related to these keywords.  Our volunteers then
attempted to access these sites from within China using our testing
application. 

Some of the most noteworthy of the topics censored include:

•    Information online related to opposition political
parties (more than 60% of Chinese-language sites tested were blocked);

•    Political content (90% of Chinese-language sites tested on The
Nine Commentaries, a critique of the Chinese Communist Party, and 82%
of sites tested with a derogatory version of Jiang Zemin’s name were
blocked);

•    The Falun Gong spiritual movement (44 – 73% of sites tested, in both English and Chinese languages);

•    The Tiananmen Square protest of June 4, 1989 (at least 48% of
Chinese-language sites tested, and 90% of sites related to the search
term “Tiananmen massacre”);

•    Independence movements in Tibet (31% of tested Chinese-language
sites), Taiwan (25% of tested Chinese-language sites), and Xinjiang
province (54% of tested Chinese-language sites); and,

•    Virtually all content on the BBC’s web properties and much of the content published online by CNN.

China has issued official statements about its efforts to limit access
to Internet pornography.  However, we found that less than 10% of sites
related to searches for the keywords “sex,” “pornography,” and “nude”
were blocked.  This imprecision, when compared either to the
effectiveness of China’s censoring of political content or to the
relatively thorough blocking of pornographic materials by states in the
Middle East, suggest that blocking pornography is nowhere near the
imperative that controlling political speech is in China.  It also
suggests that China’s war on pornography may be focused more on closing
domestic sources of pornography than on filtering foreign sites that
are providing pornographic content.

Our testing also found evidence that China tolerates considerable
overblocking – filtering of content unrelated to sensitive topics, but
located at URLs or with keywords similar to these subjects – as an
acceptable cost of achieving its goal of controlling Internet access
and publication.  China has managed over time to reduce the rate of
overblocking as its filtering technologies have improved.

Types of Communications Affected by China’s Filtering Regime.

China’s commitment to content control is revealed by the state’s
efforts to implement filtering for new methods of communication as they
become popular.  Most states that filter the Internet do an ineffective
job of blocking access to certain web sites, and stop there.

While China’s blocking of World Wide Web sites is well-known, much less
is known about the extent to which China blocks other forms of
Internet-based communications.  As Web logs (“blogs”) became popular in
2004, the state initially closed major Chinese blog service providers
until they could implement a filtering system.  When these providers
re-opened, their service included code to detect and either block or
edit posts with sensitive keywords.  Similarly, on-line discussion
forums in China include both automated filters and human Webmaster
inspections to find and remove prohibited content.  Most recently,
China moved to limit participation in university bulletin board systems
(BBS) that had featured relatively free discussion and debate on
sensitive topics.  The Chinese filtering regime also causes the
blockage, or dropping, of e-mails that include sensitive terms.  Our
testing of e-mail censorship suggests that China’s efforts in this area
are less comprehensive than for other communications methods, though
reports from the field suggest that the fear of surveillance and
blockage of e-mails is a serious issue for many activists regardless of
the precise extent of the censorship itself.

One of the most intriguing questions, as yet unanswered, is whether
emerging new technologies will make Internet filtering harder or easier
over time.  A new, emerging crop of more dynamic technologies –
centered on the fast-growing XML variant RSS, which is a means of
syndication and aggregation of online content, such as weblog entries
and news stories from major media outlets – should make filtering yet
harder for the Chinese and for other countries that seek to control the
global flow of information.  The cat-and-mouse game will continue.

The Legal Context of Filtering in China.

China’s intricate technical filtering regime is buttressed by an
equally complex series of laws and regulations that control the access
to and publication of material online.  While no single statute
specifically describes the manner in which the state will carry out its
filtering regime, a broad range of laws – including media regulation,
protections of “state secrets,” controls on Internet service providers
and Internet content providers, laws specific to cybercafés, and so
forth – provide a patchwork series of rationales and, in sum, massive
legal support for filtering by the state.  The rights afforded to
citizens as protection against filtering and surveillance, such as a
limited privacy right in the Chinese Constitution, which in other
situations might provide a counter-balance against state action on
filtering and surveillance, are not clearly stated and are likely
considered by the state to be inapplicable in this context.  For the
most part, the Chinese legal regime is not transparent, in the sense
that it does not describe the filtering regime.

Our analysis of China’s legal regime indicates a significant expansion
in the number of statutes, regulations, and regulatory bodies involved
in oversight and control of Internet access and content since 2000. 
These rules often appear to be arbitrary and are certainly
extraordinarily burdensome, such as rules that call for multiple
licensing and registration requirements imposed upon Internet content
providers. 

China’s legal system imposes liability for prohibited content on
multiple parties: the author who creates it, the service provider who
hosts it, and the end user who accesses it.  This combination of
transaction costs and broad liability has a substantial chilling effect
on on-line communication.

We are cognizant that, while we have taken great care in our legal
analysis of China’s filtering regime as it appears on the books, our
report may not describe the law as it applies on the ground.  Political
stability is clearly more important than legal justification for the
state’s actions, as a comparison of China’s filtering regime to the
corresponding legal framework demonstrates.

A Comparison of China with Other States that Filter.

Our studies have compared the Internet filtering practices of a series
of national governments in a systematic, methodologically rigorous
fashion.  A primary goal of this research is to reach useful,
substantive conclusions about the nature and extent of Internet
filtering in states that censor the Internet and to compare practices
across regions of the world.  Over the course of the next several
months, we will release a series of extensive reports that document and
provide context for Internet filtering, previously reported
anecdotally, in each of the dozen or so countries that we have studied
closely.   The new reports released to date – which document filtering
in Saudi Arabia, the United Arab Emirates, and Bahrain as well as in
China – will be followed shortly by other studies of other states in
the Middle East, East Asia, and Central Asia.

Filtering regimes – and their scope and level of effectiveness,
respectively – vary widely among the countries we have studied. 
Filtering is practiced at some level by most countries; it is best
thought of as a continuum of behavior rather than a binary, on-off
approach to content control.  Some countries employ only symbolic
filtering, and depend on legal or social pressures to constrain
content.  These states include Bahrain and Singapore, which block only
a few sites that are primarily pornographic in nature.  Other countries
demonstrate limited blocking but, because of an unsophisticated
approach to filtering, also censor large numbers of unrelated sites. 
This inadvertent filtering, known as “overblocking,” was demonstrated
by South Korea when it sought to prevent access to sites promoting
North Korea.  Finally, many countries employ a mix of commercial
software (from American companies such as Secure Computing and
Websense) to control content such as pornography and gambling while
also customizing their block lists to target prohibited political,
religious, and social content.

China, as documented in a number of studies and supported by the our
findings, institutes by far the most intricate filtering regime in the
world, with blocking occurring at multiple levels of the network and
covering content that spans a wide range of topic areas.  Though its
filtering program is widely discussed, Singapore, by contrast, blocks
access to only a small handful of sites, mostly pornographic in
nature.  Most other states that we are studying implement filtering
regimes that fall between the poles of China and Singapore, each with
significant variation from one to the next.  These filtering regimes
can be properly understood only in the political, legal, religious and
social context in which they arise. 

Conclusion.

By any means of comparison, the People’s Republic of China’s Internet
filtering regime has the greatest effect on the freedom of expression
of any filtering regime throughout the world.

The Chinese censors have a very difficult job as they try to contain
the flow of information on the Internet.  The most determined Chinese
Internet users can often elude the censors in nearly all instances. 
But the Chinese censors are head and shoulders above everyone else –
short of those who block access to the network altogether – in terms of
filtering the Internet.  Most citizens see a very different Internet in
China than citizens in other places around the world.

The Chinese Internet filtering regime grows more robust each day.  As
new information and communications technologies develop, the Chinese
censors track the technologies and determine means to control the
freedom of expression through the new media.  Filtering and efforts to
circumvent it are likely to continue into the foreseeable future. 
Though far from completely effective, China’s filtering regime achieves
a climate of self-censorship and a chilling of expression and
communications online, particularly when it comes to political dissent.

The Internet can be an extraordinarily empowering tool.  Individuals
who have never before had a voice – whether in China or anywhere else
that the network reaches – can  today project their voice to a
world-wide audience.  Seen from another vantage point, the way citizens
use the Internet is a threat to the political stability of the
governing Communist Party in China.  The state’s Internet filtering
regime is intended to mitigate this threat.

If deployed properly, the Internet can help foster active,
participatory democracies throughout the world.  Internet filtering and
surveillance, most clearly exemplified by China’s Internet filtering
regime, threaten to choke this potential. 

* * *

A complete study of Internet filtering in China, as of 2005, may be found at http://www.opennetinitiative.net/china/.