Monday, September 08, 2008

Critique of Zittrain's "The Future of the Internet and How to Stop It"

One book that the technorati have been talking about recently (ok, not so recently... it took me a while to write this article) is Jonathan Zittrain's The Future of the Internet and How to Stop It. For a book written by a co-founder of the Berkman Center and someone who is a remarkably good speaker, I found the work to be disappointing. The book's argument is not convincing and the writing seems to lack discipline, often wandering from one loosely related subject to another.

Zittrain's main point is that the security failings of generative technologies will push consumers to buy more restrictive, and supposedly safer, devices. This claim has a number of problems with it. The first is that tethered devices are not safer or more secure than generative ones -- in fact, normally the opposite is true. Compare the number of vulnerabilities in the Windows operating systems vs the number in Linux or BSD operating systems. Or bugs in Internet Explorer vs bugs in Firefox. This claim is even more dubious the more control the manufacturer has over the device: Richard Stallman points out in his response to Zittran that the iPhone's remote kill-switch makes the iPhone "designed for remote attack by Apple."

The second problem with Zittrain's principal claim is that a consumer has no incentive to prefer a non-generative device. Since non-generative devices are less secure than generative ones, any purported advantage that the non-generative device manufacturer could claim is lost. There is empirical evidence to support the belief that consumers prefer generative devices --Stallman cites the number of jailbroken iPhones as an example. Roger Grimes adds in his response: "It’s hard to say that closed systems are taking a more prominent role when open examples abound. Even the 'closed' systems he mentions are becoming more open thanks to competition and customer demand."

Even if, for the sake of argument, locked-down devices were somehow more secure than generative devices, consumers wouldn't necessarily migrate to non-generative appliances because users rarely make purchasing decisions based on security. Most computers are purchased because the user is comfortable with the platform or because he thinks that the computer is pretty or because that particular computer is necessary to run some type of software. Rarely will a run-of-the-mill consumer take into account a record of operating system vulnerabilities or the pros and cons of different systems architectures when deciding between OSX and Windows.

There are other shortcomings of the book besides the weakness of the main argument. For one, Zittrain mistakes generativity as being a zero-sum game: something is either generative or it isn't. There is a continuum of generativity: for instance, Linux is more generative than Windows XP, but Windows XP is more generative than Windows Vista. It is a fallacy to simply assume that all products fall into one non-generative bucket or the other generative one.

For a book whose title suggests solutions to the problems with the Internet, Zittran's ideas underdeliver. Virtual machines, extra-legal incentives, data portability and network neutrality are all things that are familiar, and have been, to policymakers and programmers for a while. In a book such as this which only worries about theoretical overtures and not about the detailed technical implementation, more out-of-the-box, grander thinking and proposals would have been welcome.

The book has a couple of chapters that feel decidedly out of place. The final chapter regarding privacy and the chapter exploring Wikipedia both don't seem to fit in to the framework of the book. That being said, both are certainly worthy of scholarship on their own merits. I particularly found the chapter on privacy engaging, if not particularly relevant to the rest of the book.

Zittrain's book is still worth a read: it addresses areas of concern in today's Internet and references much interesting material. The end result, however, is unconvincing and disappointing -- keep a few grains of salt handy when reading.


Jonathan Zittrain
"Protecting the Internet Without Wrecking It"

Richard Stallman
"The root of this problem is software controlled by its developer"

Bruce M Owen
"As long as flexibility has value to users, suppliers will have incentives to offer it"

Roger A Grimes
"Fixing Web insecurity requires more than a caring community"

Hal Varian
"Ultimately, the best protection is an informed buyer who demands openness"

Susan Crawford
"In the eyes of many exiting institutions, security isn't a problem -- it's an opportunity"

David D. Clark
"We need to develop a socially embedded online experience"

Jonathan Zittrain
"The best solutions don't assume a zero-sum tradeoff between security and generativity"

Coverage on BoingBoing

Ars Technica review and interview

No comments: