Designing an Internet by David D. Clark (review)

David D. Clark is one of the key scientists helping build and reshape the Internet almost from its very inception (the first nodes of ARPA-net became functional in 1969). His book is a powerful reflection on the way the network has evolved from its very start in the early 1960s until today. The focus of Clark's book is not on the Internet as we use it 24-7 or on the countless applications that continue to be added to the system, but on the very heart of the Internet, that is, the way its fundamental architecture is designed to become what it was meant to be and what it actually still is today: an information packet transport system using specific electronic protocols to transfer information from a sender to a receiver. (The analogy with the postal system is not false, in spite of all the technological differences between the two systems.) Clark stresses how wonderfully robust this fundamental structure has proven to be, for after so many years and often radical reorientations, all essential aspects, structures and functions of the first Internet are still in place. Moreover, there does not seem to exist a strong demand for a radically different Internet today. Despite all the issues that the dizzying expansion of the network has inevitably laid bare, and the many problems of which we are now all perfectly aware, the basic design is still considered, for good or bad reasons, "good enough" to be continued.

Designing an Internet—"an" Inter-net, not "the" Internet—is thus an attempt to identify, understand, analyze and if possible remediate, albeit in very speculative ways, the progressively disclosed problems of the fundamental design (for instance problems with "naming" the packets that are transferred over the system). Clark does so by emphasizing the advantages of a minimalist design, a design philosophy that sticks to the "better done than perfect" principle, while also believing that good design should only care about the fundamental aspects of the system rather than trying to implement new or other functionalities. Such a minimalist approach not only creates more robust systems, it is also imperative when one wants to cope with problems that the first generations of Internet engineers did not—that is, could not—take into account, such as security, sustainability, social demands and, yes, profit. By the way, according to Clark, one of the major reasons why the Internet proves so curiously stable is the fact that the private sector, which came to take over from the public sector in the 1980s, does not have sufficient incentives to massively invest in the fundamental design of the system (built with public money in the years before). Technology is risky business, and the risks are lower at the level atop the fundamental architecture. In other words, private enterprises are more interested in applications than in the foundational layer of packet-sending structures.

Clark is a wise man. It suffices to notice his sense of humor. He is also a brilliant scientist, as shown by his capacity to make the Internet understandable to laypersons, who are no less than his colleagues and other specialists in the field the intended audience of this book. But he is also a great writer, not only because of his stimulating didactic style but also because of his perfect sense of timely delivery. Nothing comes too early, nothing is said with delay; the author does not shy away from useful repetitions, just as he always uses the exact number of words and sentences to express his ideas—never too little or not enough—whereas the words he uses are always the right ones, with helpful indexes, definitions and acronyms at the end of the book. In short, at any page of the book the reader has the impression that they maintain a good overview of the full argument, even if certain technical details may not always be very easy to grasp.

It is the mix of great wisdom, visionary...