It’s starting to get weird out there. When WikiLeaks released classified U.S. government documents in December, it sparked several rounds of online conflict. WikiLeaks became the target of denial-of-service attacks and lost the support of its hosting and payment providers, which inspired sympathizers to counterattack, briefly bringing down the sites of MasterCard and a few other companies. Sites related to the hackers were then attacked, and mirror sites sprang up claiming to host copies of the WikiLeaks documents—although some were said to carry viruses ready to take over the machines of those who downloaded the copies, for who knows what end. Months before, an FBI official said disruption of the Internet was the greatest active risk to the U.S. “other than a weapon of mass destruction or a bomb in one of our major cities.”
Attacks on Internet sites and infrastructure, and the compromise of secure information, pose a particularly tricky problem because it is usually impossible to trace an attack back to its instigator. This “attribution problem” is so troublesome that some law-enforcement experts have called for a wholesale reworking of Internet architecture and protocols, such that every packet of data is engraved with the identity of its source. The idea is to make punishment, and therefore deterrence, possible. Unfortunately, such a reworking would also threaten what makes the Internet special, both technologically and socially.
The Internet works thanks to loose but trusted connections among its many constituent parts, with easy entry and exit for new Internet service providers or new forms of expanding access. That is not the case with, say, mobile phones, in which the telecom operator can tell which phone placed what call and to whom the phone is registered. Establishing this level of identity on the Internet is no small task, as we have seen with authoritarian regimes that have sought to limit anonymity. It would involve eliminating free and open Wi-Fi access points and other ways of sharing connections. Terminals in libraries and cybercafes would have to have verified sign-in rosters. Or worse, Internet access would have to be predicated on providing a special ID akin to a government-issued driver’s license—perhaps in the form of a USB key. No key, no bits. To be sure, this step would not stop criminals and states wanting to act covertly but would force them to invest much more to achieve the anonymity that comes so naturally today.
The price to the rest of us would also be high. The Internet’s distinct configuration may have made cyberattacks easy to launch, but it has also kindled the flame of freedom. One repressive state after another has been caught between the promise of economic advancement through abundant Internet access and the fear of empowering its citizens to express themselves freely. An Internet without the attribution problem would introduce a new issue: citizens could be readily identified and punished for their political activities.
We need better options for securing the Internet. Instead of looking primarily for top-down government intervention, we can enlist the operators and users themselves. For example, Web site operators could opt into a system of “mirror as you link.” Whenever their servers render a page, they cache the contents of the link. Then, when someone tries to get to the site and can’t, he or she can go back to the original linking site and digitally say, “I can’t get that link you just directed me to. Would you mind telling me what was there?”
Such a system of mutual aid would draw on the same cooperative and voluntary instinct behind the development of the Internet itself. If I participate as a Web site, I will know that others linking to me will also mirror my material; we each help the other, not simply because it’s the right thing to do, but because we each benefit, spreading the risk of attack and cushioning its impact among all of us. It’s a NATO for cyberspace, except it would be an alliance of Web sites instead of states.
A mutual aid framework could also make the Internet secure in other ways. PCs can alert others not to run code that just sickened them, signaling health levels to others. Internet providers could also develop technologies to validate their relationships to one another and ferret out misleading data, the way Wikipedia volunteers can quickly act to roll back thousands of acts of vandalism a day.
We rightly fear our networks and devices being attacked—but we should not let this fear cause us to destroy what makes the Internet special. We have to become more involved and more subtle—and soon.