Fouling Our Own Net

The future is far from rosy in Jonathan Zittrain's The Future of the Internet and How to Stop It

It would be easy to ignore a book called The Future of the Internet and How to Stop It . First, it predicts the progress of technology, which as every engineer knows is a risky business. Second, it seems to inveigh against one of the most successful and transformative inventions of our time. Has author Jonathan Zittrain, a cofounder of Harvard Law School’s Berkman Center for Internet & Society, suddenly become a modern Luddite, urging us to abandon the connectivity we have come to love?

Not to worry. Zittrain favors this connectivity and wants us to nurture it. The foundation of his argument is that the Internet fosters creative, collaborative invention—that it is, in a word, generative . This generativity arose in part because of its creators’ architectural principles. One such principle called for the Internet Protocol to be like the neck of an hourglass—slender yet open to all. Another principle was to move, whenever possible, all functions to host computer systems at the edge of the Internet. Together these two principles ensured that the middle of the network didn’t interfere with end-user innovations, such as better search tools or streaming media.

Zittrain considers two opposing challenges to generativity. One challenge, intrinsic to any generative network, is insecurity. Users control their computers and therefore control the network as a whole, inflicting on others any harm they please, such as spam and viruses.

The second challenge comes from outside the Internet. Carriers, service providers, and governments impose technologies that lock down our systems so that we, the end users, cannot control anything but their most trivial aspects. Zittrain calls such imposition ”perfect enforcement,” and he notes that it can be brought to bear through technical and legal constructs designed to prevent creative use or to revoke existing capabilities. One key example originated outside the Internet but is all too applicable to it—a court decision that ordered EchoStar to disable the ”record” function in its customers’ set-top boxes, a function its users had already purchased (or thought they had).

Zittrain’s most controversial and novel point concerns cloud computing, in which data and software reside on remote servers instead of securely within one’s own computer. Many would characterize cloud computing and related technologies as highly generative, but Zittrain argues that rather than liberating us from perfect enforcement, the freedom to ”mash up” services may make it even easier to lock down our data and software.

So is this the way the Internet’s generativity ends, with robotic spies and saboteurs everywhere you click? A sterile, Stepford-like sameness? Entrepreneurs fighting to control and profit from your every move?

Not necessarily. Despite these dark and powerful forces, Zittrain suggests we can save the Internet. The temptation to create schemes for perfect enforcement can and ought to be resisted, both by governmental regulators seeking policy solutions to social problems and by users seeking safety. Zittrain has no silver bullets but argues that collective action by self-organized groups of users could solve these problems. Wikipedia is one example of such a self-governing generative system.

Zittrain concludes with a speculative discussion of what he calls ”Privacy 2.0,” the challenges to personal and social privacy that arise from pervasive real-time sensing, data capture, permanent logging, inference, and dissemination. Those challenges are still in their early stages, but the costs of capturing, storing, and transmitting data are falling so fast that huge volumes of information about our personal and social activities are being archived throughout the Internet in places controlled by people and organizations who have never before had the ability to link to or analyze such information.

With such knowledge comes great power, and it is a remarkable fact about our age that this power accrues not just to 1984 -style Big Brother governments but to any group willing to share and trade information with other groups. Again we see Zittrain’s generative dilemma: The new Internet-based capabilities are as dangerous as they are fantastic, potentially transferring control from users to those who may be interested only in narrow exploitation.

Zittrain’s solution—Wikipedia-like enforcement of flexible norms under collective user control—suggests an important direction. Yet this is perhaps the point where the argument needs a systems engineer’s perspective. It is understandable that Zittrain, a law professor, sees code as something relatively substantial, concrete, and visual. He imagines lines of text laboriously typed by a user, run through a compiler, or stored in a version control system or other repository. Yet today’s code is anything but stable, substantial, and static. Modern software is evolving into systems that write code dynamically, sometimes even rewriting their own code in response to changes in the environment.

I suspect we are on the verge of a new kind of Internet-based, self-referential ”generative engineering,” which I imagine to be the emerging science of information systems that we don’t construct so much as influence—systems that construct themselves out of components in the cloud. Generative engineering is likely to present control issues of a scale we have not yet encountered. Zittrain’s book is extremely helpful in moving the debate forward, but his yearning for a future in which we ”stabilize” generativity between the extremes of the Generative Dilemma and Perfect Enforcement may be impossible to satisfy.

If this science of generative engineering takes off, the future Internet may not be stable in any way at all. How can we stop that?

About the Author

David P. Reed, an IEEE Fellow, is an adjunct professor at the MIT Media Lab. He was an early contributor to the TCP/IP protocols and designed UDP, the main alternative to TCP. He coauthored the seminal paper "End-To-End Arguments in System Design," originally presented at the 1981 IEEE Second International Conference on Distributed Computing Systems, in Paris.