who come under attack by DDoS or who introduce complex legal questions. Furthermore, when human rights activists congregate on a small subset of servers, traditional
ISP-level filtering becomes a more effective tool for censoring sensitive speech. If all
sites critical of Burmese government policy are located on a single group of servers,
that server is certain to be blocked at a national level, and is likely to come under sustained DDoS attack. By utilizing OSPs used by nonactivists, activists raise the social cost
of traditional censorship—a country that chooses to block the Blogger.com domain to
prevent access to a subset of blogs removes access from millions of uncontroversial
Web sites, alienating citizens. Individuals who were not interested in the censored
content become aware of the censorship when they can no longer access other
Blogger.com sites.
Rather than creating a subset of Web sites that protect speech, it would be vastly
better to see OSPs affirm their roles as providers of free speech tools to users throughout
the world. As discussed earlier, this is a difficult decision for an organization, particularly a for-profit company, to make in isolation. At the moment, companies seem to
be choosing a legally cautious path, disabling access for users in sanctioned countries
before experiencing pressure from activists.
The experience of the successful protest against LinkedIn’s block of Syrian users suggests that one powerful tool activists have is public protest. While companies may
make a calculated financial decision to discontinue services to certain users, public
pressure can add another factor into the equation—the potential lost business from
bad publicity. While the LinkedIn protest shows the power of this strategy, BlueHost’s
decision not to reconsider their terms of service shows that the influence of public
pressure may be limited.
Given the importance of OSPs as a space for open, public speech, it is necessary to
consider their responsibilities as common carriers. For OSPs to limit their liabilities as
common carriers, they should be required to provide services to anyone legally using
these services, even if their usage is likely to attract DDoS attacks. To do otherwise is
to allow attackers a ‘‘heckler’s veto,’’ an ability to silence speech by creating a damaging and expensive response to that speech. If OSPs are required to provide services to
any law-abiding users, an appropriate response to this form of intermediary censorship
is legal action to address discrimination, not public protest. An affirmation of OSPs’
role as common carriers would not resolve the situation Iranian and Syrian users are
facing, but it might invite legal action that would force clarification of U.S. Treasury
sanctions.
Conclusion
In countries like China, where online speech is carefully monitored and controlled, we
are likely to see intermediary censorship emerge as an increasingly important compo82
Ethan Zuckerman
ISP-level filtering becomes a more effective tool for censoring sensitive speech. If all
sites critical of Burmese government policy are located on a single group of servers,
that server is certain to be blocked at a national level, and is likely to come under sustained DDoS attack. By utilizing OSPs used by nonactivists, activists raise the social cost
of traditional censorship—a country that chooses to block the Blogger.com domain to
prevent access to a subset of blogs removes access from millions of uncontroversial
Web sites, alienating citizens. Individuals who were not interested in the censored
content become aware of the censorship when they can no longer access other
Blogger.com sites.
Rather than creating a subset of Web sites that protect speech, it would be vastly
better to see OSPs affirm their roles as providers of free speech tools to users throughout
the world. As discussed earlier, this is a difficult decision for an organization, particularly a for-profit company, to make in isolation. At the moment, companies seem to
be choosing a legally cautious path, disabling access for users in sanctioned countries
before experiencing pressure from activists.
The experience of the successful protest against LinkedIn’s block of Syrian users suggests that one powerful tool activists have is public protest. While companies may
make a calculated financial decision to discontinue services to certain users, public
pressure can add another factor into the equation—the potential lost business from
bad publicity. While the LinkedIn protest shows the power of this strategy, BlueHost’s
decision not to reconsider their terms of service shows that the influence of public
pressure may be limited.
Given the importance of OSPs as a space for open, public speech, it is necessary to
consider their responsibilities as common carriers. For OSPs to limit their liabilities as
common carriers, they should be required to provide services to anyone legally using
these services, even if their usage is likely to attract DDoS attacks. To do otherwise is
to allow attackers a ‘‘heckler’s veto,’’ an ability to silence speech by creating a damaging and expensive response to that speech. If OSPs are required to provide services to
any law-abiding users, an appropriate response to this form of intermediary censorship
is legal action to address discrimination, not public protest. An affirmation of OSPs’
role as common carriers would not resolve the situation Iranian and Syrian users are
facing, but it might invite legal action that would force clarification of U.S. Treasury
sanctions.
Conclusion
In countries like China, where online speech is carefully monitored and controlled, we
are likely to see intermediary censorship emerge as an increasingly important compo82
Ethan Zuckerman