a national filtering policy may be seen as either too high, too difficult to maintain, or simply
undesirable for other reasons.
However, the costs of no control may be even higher. In this respect, the ‘‘color revolutions’’
that occurred in the former Soviet republics of Ukraine, Georgia, and Kyrgyzstan between
2003 and 2005—which leveraged the Internet and other forms of communications as a
means to force political change by way of mass civil action—may be seen as milestones
toward the evolution of just-in-time blocking identified by the ONI. Just-in-time blocking differs
from the first-generation national filtering practices of countries like China and Iran in several
significant ways. First, and most importantly, just-in-time blocking is temporally fixed. Unlike
the evolving block lists used by national firewalls, just-in-time blocking occurs only at times
when the information being sought has a specific value or importance. Usually, this will mean
that blocking is imposed at times of political change, such as elections, or other potential social flashpoints (important anniversaries or times of social unrest). In the CIS, this kind of filtering was documented by the ONI during the March 2005 Kyrgyz parliamentary elections,
34 the
March 2006 Belarus presidential elections,
35 and the October 2006 Tajik presidential elections.
36 It has also been alleged in other regions, including Bahrain, Uganda, and Yemen during the run-up to their 2006 presidential and parliamentary elections.
37
Second, the exact techniques by which just-in-time blocking is occurring differs greatly from
traditional national firewalls. In some cases, such as in the Tajik and Ugandan elections, existing public order laws are used that require ISPs to filter out sites detrimental to national security. In Tajikistan, ISPs received orders to block two opposition Web sites ‘‘in compliance with
the national concept of information security developed in year 2003’’ as they were deemed to
‘‘aim to undermine the state’s policies in the sphere of information.’’
38 Similarly the Uganda
Communications Commission (UCC) ordered the two national Internet providers, MTN and
Uganda Telecom, to block radiokatwe.com, a Web site critical of the government citing ‘‘serious concerns.’’
39 The Uganda case came to light as the technique used by the ISP resulted in
a further 657 completely unrelated Web sites that shared the same IP address being blocked.
Bahrain blocked several Web sites in the run-up to the country’s parliamentary elections in
2006, and Yemen banned access to several media and local politics sites ahead of the country’s 2006 presidential elections. Likewise, Bahrain also briefly blocked access to Google
Earth in 2006, citing national security concerns, as did Jordan in the same year with respect
to Skype.
In other cases, blocking has been accomplished by covert or special technical means. During the Belarus elections, a variety of techniques were observed, ranging from apparent errors
in the propagation of domain name information, causing Web sites to be inaccessible from
ISPs within Belarus, through to technical failures that disconnected all Internet access in Minsk
during the period of street demonstrations that followed the election. One of the most often
seen techniques is the use of Denial-of-Service attacks (DoS) against ISPs hosting targeted
Web sites or services. This form of blocking is particularly effective as it can occur anony144
Ronald Deibert and Rafal Rohozinski
Précédent

- 161/467

Suivant