constitute censorship. It seems likely that we will see political rivals attempt to disable
each other’s online speech using spurious copyright claims, even though these claims
run the risk of exposing a complainant to penalties for acting in bad faith.
We are already beginning to see attacks on online speech that attempt to trigger an
‘‘immune system response’’ from hosting companies. Irrawaddy, a leading Web site for
Burmese dissidents, suffered a series of distributed denial of service (DDoS) attacks in
September 2008. Other pro-dissident sites, including the Oslo-based Democratic Voice
of Burma and the New Era Journal, based in Bangkok, were rendered inaccessible by
DDoS attacks, prompting speculation that the attacks were the work of hackers hired
by the Burmese government. 35 As Irrawaddy struggled to fend off the DDoS attacks,
its hosting provider became increasingly agitated, since the DDoS attack affected their
other customers. For a period of time, the attack was so severe that Thailand’s primary
Internet connection was overloaded with DDoS traffic. The attack had the immediate
effect of making the site unreachable, as well as the longer-term effect of forcing Irrawaddy to find a new ISP. 36
As with copyright takedown notices, OSPs have a strong incentive to remove ‘‘troublesome’’ users—DDoS attacks can require hours of expensive system administration
time to fend off. For hosting accounts that generate little revenue, the cost of fending
off even a small DDoS is likely to exceed profit margins, and OSPs have an incentive to
remove customers who have come under attack. Providers that develop a reputation
for protecting their users from DDoS attacks are likely to attract customers who come
under DDoS attacks, increasing their costs. As such, some OSPs are comfortable removing customers because they have come under attack. A libel lawsuit between Colocation America and Archie Garga-Richardson, whose Web site came under DDoS attack
while hosted by Colocation America, makes it clear that OSPs are sometimes willing
to remove customers generating thousands of dollars of revenue. 37
Implications for Researchers
While the underlying threats to speech are deeply different between the case of
Chinese blogging hosts studied by MacKinnon and U.S.-based OSPs outlined here, a
common theme emerges: threats to online speech come not just from government
action, but from the needs of OSPs to interpret and follow government regulations
and to turn an operating profit. If this trend persists and grows, it has implications for
scholars and activists focused on this issue.
Scholars need to develop tools and methods to study corporate filtering of Internet
content. These methods might incorporate the techniques pioneered by MacKinnon
in testing Chinese Web sites for automated censorship, perhaps with development of
more robust tools to help automate testing. They could also include a site like Chilling
Effects focused on collecting reports of corporate censorship of content, or an expan80
Ethan Zuckerman
each other’s online speech using spurious copyright claims, even though these claims
run the risk of exposing a complainant to penalties for acting in bad faith.
We are already beginning to see attacks on online speech that attempt to trigger an
‘‘immune system response’’ from hosting companies. Irrawaddy, a leading Web site for
Burmese dissidents, suffered a series of distributed denial of service (DDoS) attacks in
September 2008. Other pro-dissident sites, including the Oslo-based Democratic Voice
of Burma and the New Era Journal, based in Bangkok, were rendered inaccessible by
DDoS attacks, prompting speculation that the attacks were the work of hackers hired
by the Burmese government. 35 As Irrawaddy struggled to fend off the DDoS attacks,
its hosting provider became increasingly agitated, since the DDoS attack affected their
other customers. For a period of time, the attack was so severe that Thailand’s primary
Internet connection was overloaded with DDoS traffic. The attack had the immediate
effect of making the site unreachable, as well as the longer-term effect of forcing Irrawaddy to find a new ISP. 36
As with copyright takedown notices, OSPs have a strong incentive to remove ‘‘troublesome’’ users—DDoS attacks can require hours of expensive system administration
time to fend off. For hosting accounts that generate little revenue, the cost of fending
off even a small DDoS is likely to exceed profit margins, and OSPs have an incentive to
remove customers who have come under attack. Providers that develop a reputation
for protecting their users from DDoS attacks are likely to attract customers who come
under DDoS attacks, increasing their costs. As such, some OSPs are comfortable removing customers because they have come under attack. A libel lawsuit between Colocation America and Archie Garga-Richardson, whose Web site came under DDoS attack
while hosted by Colocation America, makes it clear that OSPs are sometimes willing
to remove customers generating thousands of dollars of revenue. 37
Implications for Researchers
While the underlying threats to speech are deeply different between the case of
Chinese blogging hosts studied by MacKinnon and U.S.-based OSPs outlined here, a
common theme emerges: threats to online speech come not just from government
action, but from the needs of OSPs to interpret and follow government regulations
and to turn an operating profit. If this trend persists and grows, it has implications for
scholars and activists focused on this issue.
Scholars need to develop tools and methods to study corporate filtering of Internet
content. These methods might incorporate the techniques pioneered by MacKinnon
in testing Chinese Web sites for automated censorship, perhaps with development of
more robust tools to help automate testing. They could also include a site like Chilling
Effects focused on collecting reports of corporate censorship of content, or an expan80
Ethan Zuckerman