"Defining Tomorrow's Internet" Review

The Internet was born as a research network, designed and extended by
parties who shared the common goal of interconnecting the world's
computers so that new types of applications could be invented. As the
Internet's influence over mainstream society grew, its political,
social, and commercial value increased rapidly, creating diverse
groups of stakeholders who sometimes pursue opposing goals. These
stakeholders are engaged in ongoing struggles (termed "tussles") to
change and extend the Internet according to their interests.
According to the authors, network architecture decisions should be
made so that these tussles can play out without affecting unrelated
parts of the architecture. The paper introduces a new way of thinking
about the changes happening to the Internet architecture and describes
a number of technical design principles that could be used to
accommodate tussles.

The main recommendation of the paper is to design for variation in
outcome, such that outcomes can be different in different places and
so that the tussles can take place within the original design. This
principle is further refined into "modularize the design along tussle
boundaries" and to "design for choice" so that different stakeholders
may express their preferences with the choices they make. These
principles have additional, more specific implications: choice often
requires open interfaces, visibility of the consequences of choice
matters (otherwise it can be ignored and it becomes insignificant),
tussles and the weapons used to fight them will evolve over time,
there is no value-neutral design, and designers should not assume that
their design solves a struggle, instead they should only provide a
playing field. A section of the paper is devoted to explaining how
these concepts are embodied in current struggles and recommendations
are made as to how the design could have been improved.

Overall, the paper was enjoyable to read, I believe the authors have
an original and profound insight into the major struggles shaping the
Internet. The discussion of the posterity of the end-to-end argument
and the new assumptions about the future of the Internet was
enlightening. The paper does occasionally falter when explaining
certain illustrating examples and possible solutions, such as the
arguments for source-based routing and ISP-independent IPs. Although
theoretically sound, it seems unrealistic that network designers would
apply the authors' principles when solving these problems.
Received on Mon Nov 27 2006 - 22:58:28 EST