Data Casts Doubt on Glanz and Markoff's Egypt Kill Switch Theory

02/16/2011 05:19 pm ET | Updated May 25, 2011

In theNew York Times, reporters James Glanz and John Markoff investigate the actions that led to Egypt's Internet shutdown from Jan 27 - Feb 2.  They argue that the primary targets of the Egyptian Government were the international fiber optic connections that connect Egypt with the rest of the world and that were directly under the government's control.  They also claim that the generalized disruption to Egypt's domestic internet service was an unintended side effect of the county's reliance on servers outside of Egypt to provide crucial support for domestic traffic.

A closer look at the shutdown raises doubts about Glanz and Markoff's theory of how the Egyptian government was able to accomplish the disruption.  For starters, the suggestion that the Mubarak regime was more interested in cutting Egyptians' connection to the outside world than in disrupting the organizing potential of the internet within Egypt, a step that would presumably discourage protest in the streets, is dubious.  From a purely technical standpoint, understanding the precise sequence of events that led to the shutdown will be critical to preventing other authoritarian governments from similar acts in the future.  Separately, by focusing on the centralized nature of international links, the article places the majority of the blame on the Egyptian government and overlooks the role of private enterprise in enabling this shutdown.

The available data from the time of the disconnection, and later when services were restored, do not support the reporters' “kill switch” theory.  Specifically, as demonstrated in this graphic and this PowerPoint presentation (both available on the Times' website), it is clear that the services were shut down in a relatively short window, but that the actual disconnections were handled on the ISP level. They did not occur in one fell swoop.  Each of Egypt's major internet service providers went down separately and independently of the other rather than as the result of a single action at a centralized routing location.

In addition, the shutdown of Egypt's external connectivity was not total.  One ISP, Noor Group, managed to stay up for three days after the general termination of service, but was eventually shut down on January 31st.  Noor is a small ISP that leases space on major connections from Egypt's larger service providers.  If the disconnection had been based on those connections alone, Noor would not have been able to continue offering service once international connections had been severed.  At this time there is a lot of speculation, but little good information, on why and how Noor stayed up when all other connections went down.  However, the example of the Noor Group suggests that there may have been room for independent action on the part of Egyptian service providers to choose to remain up and offer service.

The evidence for how Egypt's connection to the internet was shut down is fairly clear.  On January 27 between 10:15PM GMT and 10:35PM GMT, Egypt's main internet service providers removed approximately 90% of their routes for what is known as the Border Gateway Protocol (BGP).  BGP is the method by which networking hardware connects to independent smaller networks, known as autonomous systems (AS). With the BGP routes blocked, Egypt's citizens went from being connected to the entire world wide web to connecting only on 50+ smaller networks that were unable to see each other or anyone outside of Egypt.  Between 9:30 GMT and 10:00 GMT on the morning of Febuary 2 the internet service providers resumed broadcasting on the original BGP routes and full connectivity was quickly restored.

It is not surprising that some connectivity remained during the shutdown period.  As described by Glanz and Markoff, connections within a single AS would have been unaffected by the shutdown.  For example, users might well have been able to achieve limited access to a few websites hosted by the same organization that was providing them with internet access.

It's crucial that we understand the  technical story of what happened in Egypt, because following the connections leads to an important - and not fully appreciated -- aspect of the Egyptian uprising. The evidence suggests that the Egyptian government would have required the cooperation of private companies in order to accomplish its shut down.

This is why Human Rights First sent letters to Egypt's 10 largest internet service providers.   In these letters we asked:

  • How did your company decide to terminate its ISP services?
  • Did you notify your Egyptian users before terminating service and if not, do you have any plans to communicate your decision publicly?
  • If you came to the view that you had no legal option but to comply with an official order, what was the reasoning behind that decision?
  • Did your company seek additional information or clarification from the Egyptian government?

So far, only one company has responded to our request for information.