Complexity: A Guided Tour

  • Melanie Mitchell
Oxford University Press: 2009. 368 pp. £14.99, $29.95 9780195124415 | ISBN: 978-0-1951-2441-5

For several decades, scientists studying complex systems — rich, collective systems such as ant colonies, economies and cells — have spoken of 'emergence', the mysterious process by which the collective whole acquires resilience, adaptability and other surprising properties, even though its components are simple. The archetypal example is the ant colony, which manages to forage intelligently for food and organize collective defence by exploiting the limited skills of its individual ant citizens.

As a colony, ants perform complex tasks that individuals could not achieve alone, such as tending larvae. Credit: MEUL/ARCO/NATURE PICTURE LIBRARY

The ideas of complexity have spread across science, and emergence has become a buzzword. Philosopher Mark Bedau has suggested, however, that it poses a puzzle, as it demands that two seemingly contradictory statements must be true. In complex systems, organized phenomena at higher levels depend on processes at lower levels: everything in a cell, for example, depends on the processes of atomic physics. Yet phenomena emerging at higher levels gain autonomy from lower levels: the body's organs and their interactions can be described and explained without reference to atomic physics.

How can something be dependent and autonomous at the same time? And why do so many systems in nature show this hierarchical organization? No one has answered these questions, but in Complexity, computer scientist Melanie Mitchell of the Santa Fe Institute, New Mexico, offers a valuable snapshot of the growing field of complex-systems science from which the answers may eventually arise.

Mitchell explores the historical roots of this area in the work of visionaries such as Henri Poincaré and Edward Lorenz in dynamical-systems theory, and of John von Neumann, Alan Turing and others in computation. The unifying feature of complex systems, Mitchell emphasizes, is that systems as diverse as cells, economies and ecosystems, as well as the human brain, all process information, and do so in a way that makes them rich, adaptable and hard to understand. The book hits its stride in its latter half, with an insightful survey of recent developments in complex-network theory and scaling in biology.

Especially valuable is the book's exploration of recent attempts to categorize the dynamics of cellular automata — simple systems that act as models for the study of rich dynamics. Some of this work, under the name of computational mechanics and linked to the ideas of Mitchell's former colleague, the late Jim Crutchfield, probes the fundamental 'information physics' of complex systems in general. This focus of the book is commendable, as much of the literature of complex-systems research dwells on more expansive philosophical themes at the expense of the 'boring' details of specific models. Yet intense scrutiny of such models may ultimately reveal clues to solving Bedau's mystery.

Mitchell touches on the many practical applications of this science, ideas put into practice by forward-looking companies such as Cisco and Capital One. The book is timely, given that many analyses of the present financial crisis have concluded that the key issue is how markets have outstripped our ability to understand them.

It has become fashionable in recent years to criticize complex-systems science for generating too much hype and not offering enough practical insight. But insights into truly complex problems do not come easy. Mitchell's welcome book makes it clear that this field is making steady, if slow, progress.