Paper review: The Design Philosophy of the DARPA Internet Protocols

Reviewer: Robert Dugas

The primary purpose for Clark’s 1988 paper on the philosophy of internet protocols is to explain the motivation behind the design decisions which led to the Internet of today. If there is a problem the author intends to solve, it is misperception of the origins of modern internetworking protocols. Clark covers a lot of ground and continually focuses on explaining the original goals and environment under which the designers of the protocols operated.

In explaining the evolution of the Internet, Clark begins by enumerating the prioritization scheme underlying the initial development of the internet. He then uses this as a jumping off point to further detail the implications of this scheme. Of primary importance and underlying even this prioritization, we are told, is the goal of connecting existing networks. This constrains even the stated top-level priority, namely survivability. The datagram building block scheme is presented as a fundamental consequence of both of these goals. An additional discussion is provided detailing the divorce of TCP from IP in order to accomplish additional requirements of service and architecture interoperability and the consequences thereof. A more general point of this paper is to present the notion that the modern internet is not any kind of holy grail or model of perfection but is instead a product of countless conflicting priorities and compromises like any major engineering project.

Valuing the significance of this paper is a philosophical issue. Clark’s goal and methodology seem primarily academic-that is to provide an accurate historical account of the designs underlying today’s internet. This information is not directly actionable in the sense of solving a problem or directing future research; the value of an historical understanding is less tangible but not necessarily lower. The reader is expected to gain insight into the factors which influenced today’s technology and use that insight to guide future contributions.