Following a recommendation of Brian Klais, I've been reading Albert-Laszlo Barabasi's Linked: How Everything Is Connected to Everything Else and What It Means.
The quick summary: the old model of random networks, going back to Erdos, doesn't tell us much about the networks that dominate our lives: the web, social nets, biochemical nets, etc. New models based on network growth do. These new models explain alot: why we see power laws in networks; how early hubs like Yahoo have a tremendous advantage in gaining links and promonience; how late entrants like Google can still sieze share late in the game; how scale-free networks are simultaneously robust to random damage but critically vulnerable to malevolent attack; how AIDS spread; why certain hub genes are involved in a disproportionately large number of phenotypes, etc.
Barbasi isn't a popular writer interpreting science (eg Gladwell or Surowiecki) -- he's a real mathematician responsible for some of this theory. And he writes incredibly well, applying his research to larger issues of technology, security, biology, sociology, and policy.
From the Amazon review:
How is the human brain like the AIDS epidemic? Ask physicist Albert-László Barabási and he'll explain them both in terms of networks of individual nodes connected via complex but understandable relationships. Linked: The New Science of Networks is his bright, accessible guide to the fundamentals underlying neurology, epidemiology, Internet traffic, and many other fields united by complexity.
Barabási's gift for concrete, nonmathematical explanations and penchant for eccentric humor would make the book thoroughly enjoyable even if the content weren't engaging. But the results of Barabási's research into the behavior of networks are deeply compelling.