confidence, perhaps not well-placed, that such self-identification would not put the user in
jeopardy of state sanction.
Trends in Internet Filtering
Researchers associated with the OpenNet Initiative have been collecting empirical data on
Internet filtering since 2001. Our methodology circa 2006 is far more sophisticated than it
has been in the past. The coverage of our research is far broader, now covering every state
known or credibly suspected to carry out Internet filtering. During this five-year period, we have
observed the following trends:
The overall trend in Internet filtering is toward more states adopting filtering regimes. The
states with the most extensive filtering practices fall primarily in three regions: east Asia,
the Middle East and North Africa, and central Asia. State-mandated, technical filtering
does occur in other parts of the world, but in a more limited fashion, such as the Internet
filtering common in libraries and schools in the United States, child pornography filtering
systems in northern Europe, and the filtering of Nazi paraphernalia and Holocaust denial
sites in France and Germany.
Some of the newest filtering regimes, such as those coming online in the Commonwealth
of Independent States (CIS), appear to be more sophisticated than the first-generation
systems still in place in some states. The early means of filtering—such as Saudi Arabia’s early system, with a heavy emphasis on pornography and offering citizens a clear
blockpage—are no longer the only ways to accomplish Internet filtering. The net result is
greater variation in what it means to filter content online.
In the Commonwealth of Independent States and in parts of the Middle East and North
Africa, the filtering we are seeing is highly targeted in nature and carried out ‘‘just-intime’’ to block access to information during sensitive time periods. ONI principal investigator Rafal Rohozinski and his coauthor Deirdre Collings predicted such an eventuality:
‘‘In democratically-challenged countries, we are likely to see increasing constraints on
the ‘openness’ of the Internet during election periods, and these constraints may be
more subtle than outright filtering and blocking.’’
25
The ONI has monitored three elections to date, one in Kyrgyzstan (2005), one in Belarus
(2006), and one in Nigeria (2007). As Rohozinski and Collings wrote,
The February 2005 elections in Kyrgyzstan marked the ONI’s first foray into election
monitoring. During the Kyrgyz elections ONI researchers were able to document two
major Denial of Service (DoS) attacks directed against ISPs hosting major opposition
newspapers. The attacks were commissioned from a commercial ‘‘bot herder’’ and
traced back to a group of Ukrainian hackers-for-hire. ONI was not able to identify who
was ultimately responsible for these attacks. Direct links to the Kyrgyz authorities could
not be established. Thus, while no direct filtering took place, the DoS attack resulted in
the indirect censorship of websites while exonerating the Kyrgyz authorities of any direct responsibility. The Kyrgyz case also raised the issue of who benefits most from
this kind of indirect filtering. In Kyrgyzstan, the target of the DoS attacks—opposition
Internet Filtering: The Politics and Mechanisms of Control
41
jeopardy of state sanction.
Trends in Internet Filtering
Researchers associated with the OpenNet Initiative have been collecting empirical data on
Internet filtering since 2001. Our methodology circa 2006 is far more sophisticated than it
has been in the past. The coverage of our research is far broader, now covering every state
known or credibly suspected to carry out Internet filtering. During this five-year period, we have
observed the following trends:
The overall trend in Internet filtering is toward more states adopting filtering regimes. The
states with the most extensive filtering practices fall primarily in three regions: east Asia,
the Middle East and North Africa, and central Asia. State-mandated, technical filtering
does occur in other parts of the world, but in a more limited fashion, such as the Internet
filtering common in libraries and schools in the United States, child pornography filtering
systems in northern Europe, and the filtering of Nazi paraphernalia and Holocaust denial
sites in France and Germany.
Some of the newest filtering regimes, such as those coming online in the Commonwealth
of Independent States (CIS), appear to be more sophisticated than the first-generation
systems still in place in some states. The early means of filtering—such as Saudi Arabia’s early system, with a heavy emphasis on pornography and offering citizens a clear
blockpage—are no longer the only ways to accomplish Internet filtering. The net result is
greater variation in what it means to filter content online.
In the Commonwealth of Independent States and in parts of the Middle East and North
Africa, the filtering we are seeing is highly targeted in nature and carried out ‘‘just-intime’’ to block access to information during sensitive time periods. ONI principal investigator Rafal Rohozinski and his coauthor Deirdre Collings predicted such an eventuality:
‘‘In democratically-challenged countries, we are likely to see increasing constraints on
the ‘openness’ of the Internet during election periods, and these constraints may be
more subtle than outright filtering and blocking.’’
25
The ONI has monitored three elections to date, one in Kyrgyzstan (2005), one in Belarus
(2006), and one in Nigeria (2007). As Rohozinski and Collings wrote,
The February 2005 elections in Kyrgyzstan marked the ONI’s first foray into election
monitoring. During the Kyrgyz elections ONI researchers were able to document two
major Denial of Service (DoS) attacks directed against ISPs hosting major opposition
newspapers. The attacks were commissioned from a commercial ‘‘bot herder’’ and
traced back to a group of Ukrainian hackers-for-hire. ONI was not able to identify who
was ultimately responsible for these attacks. Direct links to the Kyrgyz authorities could
not be established. Thus, while no direct filtering took place, the DoS attack resulted in
the indirect censorship of websites while exonerating the Kyrgyz authorities of any direct responsibility. The Kyrgyz case also raised the issue of who benefits most from
this kind of indirect filtering. In Kyrgyzstan, the target of the DoS attacks—opposition
Internet Filtering: The Politics and Mechanisms of Control
41