- Real-Time Testing of Internet Filtering in China
Documentation of Internet Filtering Worldwide
Zittrain and Benjamin Edelman
Berkman Center for Internet & Society
Harvard Law School
What is the purpose of this system?
We study filtering of the Internet, hoping to understand and document what forces shape communications on the Internet and how they do so. Certain national governments restrict Internet usage by prohibiting and mechanically preventing access to certain Web sites; China is among the governments that does so, and the testing system provided here allows interested Internet users to get a sense of what Web sites China filters.
Why the need for this special system? Why not just ask China what sites they block?
In many instances, filtering of the Internet is opaque. The authors know of no complete list of sites blocked in China, nor even of any document or Web page authoritatively and officially describing the Chinese government's filtering policy. This is typical -- even the American companies that design and sell commercial filtering applications do not publish their block lists (with some exceptions such as Netnanny); one author is currently in litigation to attempt to obtain one such list from filtering company N2H2.
So how do I use the site?
Just fill out the form on the main "Real-Time Testing of Internet Filtering in China" page. You can provide a URL you'd like to test (example: http://www.yahoo.com), then fill out the rest of the form and press submit. Wait 15-120 seconds for testing to take place, then you'll see the results.
Do I need to test every page on a site in question? Maybe China is blocking some of the pages but not others?
In general, you need only test a single URL per web site. (We suggest the default page.) While some filtering systems are capable of filtering on the level of an individual page or individual directory, China's system appears not to do so. Instead, available data strongly suggests that China's system filters on the level of an IP address. Thus, if one page of a site is blocked (or if one site on a shared "virtual server" is blocked), then the rest of that page (and any other "virtual servers" on that physical server) must also be blocked, so long as the entire site is hosted from a server or servers answering from the same IP address.
Of course, like every rule, this one has an exception: In some parts of China, filtering recently (in September 2002) began to include certain prohibitions on the use of certain keywords and phrases, whether in URLs or in HTML page text. When one page on a site includes such terms, that page may be blocked -- even as the rest of the site remains available as usual. The authors are currently attempting to prepare a list of the words and phrases filtered in this way.
How does this testing system work, anyway, and how reliable is it?
We've used a variety of methods to seek to access Web sites the same way computers in China do, and to record the results. Our methods have been refined over time, and we will document them fully in our upcoming report on Chinese filtering. For the moment, a report from the real-time testing system that a site is inaccessible can be thought of as highly likely to indicate blocking across some if not all of China, but by no means conclusive. The inaccessibility might be due to network congestion, routing problems, or other non-filtering-related disruptions to connectivity.
Filtering in China may vary across geographic locations. Current testing using this system reflects accessibility from Beijing and Yunnan province (in an automatic rotation). Testing conducted from elsewhere may yield different results. (Certain prior testing using this system reflected accesibility from a network located in Guandgong Province (in southern China), from Beijing, again from Shanghai, from Jiangsu Province (southeast of Beijing), from Hubei province (in central China), from Heilongjaing Province (in northeast China), again from Hubei province (in central China), and from Beijing/Shanghai/Fujian Province/Hubei Province/Jiangsu Province in an automatic rotation.)
Are the results current?
Yes. All results reported via the real-time testing system reflect tests within the prior 96 hours. We'd prefer to give current results that are always freshly collected precisely when requested. However, to reduce testing load, the testing system has been modified to report recent results (retrieved within the past 96 hours) rather than retesting a URL for which recent results are already available.
Why do I get "indeterminate" answers ?
Internet filtering staff in China seem to have become aware of this system, and they often take steps to block its operation. In response, we have taken countermeasures to attempt to remain operational. At the moment, we often find ourselves in a sort of limbo in which some tests yield affirmative results one way or the other (that a given site is accessible or inaccessible from China) while other tests yield indeterminate answers that don't let us draw a conclusion as to the site's status. In this latter situation, we report a result of "indeterminate." But the testing system does work -- scores of users continue to get results every day -- so if you receive an "indeterminate" response, please wait a few minutes and then try again. We apologize for the inconvenience and are doing all we can to improve the situation.
What other work are you doing about Internet filtering in China?
The authors are currently preparing a comprehensive study of Internet filtering in China incorporating the results of testing many thousands of sites, and will document several thousand that are blocked. If you provide your email address on the "Real-Time Testing" page, we'll be sure to let you know when results are available.
What other work are you doing about Internet filtering in other countries?
The authors continue to plan studies of Internet filtering in other countries, following the framework of our prior work on filtering in Saudi Arabia. We recently released Empirical Analysis of Internet Filtering in China. We're currently thinking about the Singapore, the UAE, and Vietnam.
If you know of a country that filters the Internet, we'd be interested to hear from you.
What other work are you doing about Internet filtering more generally?
We continue to think about Internet filtering in libraries and public schools as well as homes, businesses, and elsewhere. Throughout, accuracy and effectiveness of filtering remains a major focal point of our research -- quantifying both overblocking and underblocking, relative to filtering systems' stated definitions as well as relative to constitutional and other requirements.
Can I build a robot to use your system's front end to conduct my own tests of Internet filtering?
Unfortunately not. Such usage of our system would overload it. If you'd like to test a large number of sites, please send us a note.
Last Updated: January 22, 2003