We Already Have an Internet 'Kill Switch'
Killing the Internet is easy. Or, more specifically, killing specific portions of the Internet is easy. Indeed, government intervention in Egypt has demonstrated that when push comes to shove, taking control of a nation's Internet access is perhaps easier than patrolling a physical border.
In an orchestrated move, the Egyptian government has effectively shut off both ingress and egress to the Internet. Analysis of the event has shown that only around 8 percent of normal activity continued, allowing a narrow gateway for essential activities conducted through a solitary ISP.
By this action, the Egyptian administration has catapulted itself into an exclusive club whose other members include well-known administrations, such as those in Burma and Nepal, where privacy and other rights are not on the agenda.
Intervention is not that difficult technically. No "kill switch" is needed. All that's required is a means to scale Internet blocking.
Some governments quite openly embrace technology that enables intervention. China employs its "Great Firewall," while Australia has the hotly debated "Refused Classification" (RC) program. Both of these countries manage intervention through block listing to filter out anything that they consider to be undesirable. This can be done through "scraping" or automatically scanning HTTP headers on Web pages for key words and discarding any objectionable material. Application of the scraping is done at the ISP level.
Intervention doesn't need to be this subtle, though, as the Egypt situation demonstrates. Disrupting the flow of traffic through a key routing protocol is far more effective -- and destructive.
Controlling the routing protocol for the Internet BGP (Border Gateway Protocol), the means used to exchange information between ISPs either at a national or regional level, is all that is required. Further, whoever controls the DNS (domain name server) within BGP basically controls Internet access for that domain. Indeed, many countries now enact various forms of DNS and content filtering under the guise of law enforcement and national security, with ISPs and hosts complying. An example of how vital BGP is to route functionality was evidenced by a BGP experiment that went wrong last summer, causing a "significant percentage of global Internet traffic" to be disrupted.
Leaving aside the question of privacy and human rights, governments have the technical means to control traffic flows before an ISP even gets traffic or sends it out, simply by controlling a "data valve" on the service backbone, i.e., upstream of an ISP.
The US administration is keen to play down the so-called "kill switch legislation," saying that it will give government the means to work collaboratively with the private sector to protect vital infrastructures in the event of a cyber-emergency. Meanwhile, a recent OECD report is skeptical about whether will ever be a "true cyberwar."
So as we have seen, as simply demonstrated by the Egyptian government, most governments can and do simply control and throttle the Internet via upstream and backbone data valves. So the "kill switch" is actually operational, but in the form of a variable "data control valve." The question we may now ask is who has their hands on the valve?
There is one positive advantage to the "control valve." In practice, the possibility of a real cyberwar occurring is diminishing, and OECD's verdict is reasonable.