Box 1.1
Identifying and documenting Internet filtering
Measuring and describing Internet filtering defies simple metrics. Ideally, we would like
to know how Internet censorship reduces the availability of information, how it hampers
the development of online communities, and how it inhibits the ability of civic groups to
monitor and report on the activities of the government, as these answers impact governance and ultimately economic growth. However, this is much easier to conceptualize
at an abstract level than to measure empirically. Even if we were able to identify all the
Web sites that have been put out of reach due to government action, the impact of
blocking access to each Web site is far from obvious, particularly in this networked
world where information has a habit of propagating itself and reappearing in multiple
locations. Nevertheless, every obstacle thrown into the path of citizens seeking out information bears a cost or, depending on how one views the contribution of a particular
Web site to society, a benefit. With this recognition of the inherent complexity of evaluating Internet censorship, we set out with modest goals—to identify and document
filtering.
Two lists of Web sites are checked in each of the countries tested: a global list and a
local list. The global list is a standardized list of Web sites that cover the categories
listed in table 1.1. The global list of Web sites is comprised principally of internationally
relevant Web sites with English content. The same global list is checked in each of the
countries in which we have tested. A separate local list is created for each of the countries tested; it includes Web sites related to the specific issues and context of the study
country.
These testing lists encompass a wide variety of content including political topics such
as human rights, political commentary and news, religion, health and sex education,
and Web sites sponsored by separatists and militant organizations. Pornography, gambling, drugs, and alcohol are also represented in the testing lists. The lists embody portions of the Web space that would be subject to Internet filtering in each of the countries
being tested. They are designed to unearth filtering and blocking behavior where it
exists. Background research is focused on finding sites that are likely to be blocked. In
countries where Internet censorship has been reported, the lists include those sites that
were alleged to have been blocked. These are not intended to be exhaustive lists of the
relevant subject matter, nor do we presume to have identified all the Web sites that are
subject to blocking.
The actual tests are run from within each country using software specifically designed
for this purpose. Where appropriate, the tests are run from different locations to capture
the differences in blocking behavior across Internet service providers (ISPs). The tests
are run across multiple days and weeks to control for normal connectivity problems.
The completion of the initial accessibility testing is just the first step in the evaluation
process. Additional diagnostic work is required to separate normal connectivity errors
from intentional tampering. As described in further detail later, there are a number of
technical alternatives for filtering the Internet, some of which are relatively easy to discover. Others are difficult to detect and require extensive diagnostic work to confirm.
Measuring Global Internet Filtering
11
Identifying and documenting Internet filtering
Measuring and describing Internet filtering defies simple metrics. Ideally, we would like
to know how Internet censorship reduces the availability of information, how it hampers
the development of online communities, and how it inhibits the ability of civic groups to
monitor and report on the activities of the government, as these answers impact governance and ultimately economic growth. However, this is much easier to conceptualize
at an abstract level than to measure empirically. Even if we were able to identify all the
Web sites that have been put out of reach due to government action, the impact of
blocking access to each Web site is far from obvious, particularly in this networked
world where information has a habit of propagating itself and reappearing in multiple
locations. Nevertheless, every obstacle thrown into the path of citizens seeking out information bears a cost or, depending on how one views the contribution of a particular
Web site to society, a benefit. With this recognition of the inherent complexity of evaluating Internet censorship, we set out with modest goals—to identify and document
filtering.
Two lists of Web sites are checked in each of the countries tested: a global list and a
local list. The global list is a standardized list of Web sites that cover the categories
listed in table 1.1. The global list of Web sites is comprised principally of internationally
relevant Web sites with English content. The same global list is checked in each of the
countries in which we have tested. A separate local list is created for each of the countries tested; it includes Web sites related to the specific issues and context of the study
country.
These testing lists encompass a wide variety of content including political topics such
as human rights, political commentary and news, religion, health and sex education,
and Web sites sponsored by separatists and militant organizations. Pornography, gambling, drugs, and alcohol are also represented in the testing lists. The lists embody portions of the Web space that would be subject to Internet filtering in each of the countries
being tested. They are designed to unearth filtering and blocking behavior where it
exists. Background research is focused on finding sites that are likely to be blocked. In
countries where Internet censorship has been reported, the lists include those sites that
were alleged to have been blocked. These are not intended to be exhaustive lists of the
relevant subject matter, nor do we presume to have identified all the Web sites that are
subject to blocking.
The actual tests are run from within each country using software specifically designed
for this purpose. Where appropriate, the tests are run from different locations to capture
the differences in blocking behavior across Internet service providers (ISPs). The tests
are run across multiple days and weeks to control for normal connectivity problems.
The completion of the initial accessibility testing is just the first step in the evaluation
process. Additional diagnostic work is required to separate normal connectivity errors
from intentional tampering. As described in further detail later, there are a number of
technical alternatives for filtering the Internet, some of which are relatively easy to discover. Others are difficult to detect and require extensive diagnostic work to confirm.
Measuring Global Internet Filtering
11