Kelly provides a popular survey of the state of research in artificial life and emergent behavior, with a liberal dose of projection about how this technology will be applied in daily life. The book posits that tomorrow’s technology will be grown rather than manufactured, using principles reverse-engineered from the natural world. This emphasis on potential applications of artificial life differentiates Kelly’s book from other reviews of artificial life research. Kelly seeks to brief the reader not only on what is being learned in the laboratory, but also on general principles, transcending any single research project, that can be extrapolated to engineered systems.
One difference between living systems and conventional artificial ones is that life is messy and tangled, as obscure in its detailed structure and logic as it is robust in its overall functioning. Kelly’s book emulates this approach to wholeness. It consists of 24 chapters with no explicit higher-level groupings. General themes emerge from groups of chapters, but in such a way that the reader feels them before seeing them. The book is brutally verbal, with no diagrams or graphics other than a ubiquitous Celtic logo of the tail-eating serpent Ouroboros. The bibliography lists about 300 items, each with helpful annotation, but many research reports mentioned in the text are not fully cited.
Chapter 1, “The Made and the Born,” sets forth the vision of infusing machines with the logic of life, in which order arises from the chaos of multiple interacting entities. The next two chapters exemplify this dynamic, first in terms of information processing in chapter 2, “Hive Mind,” then extended to physical domains in chapter 3, “Machines with an Attitude.”
The next six chapters develop the notion of an ecology, in which different interacting entities mutually depend on one another. Chapter 4, “Assembling Complexity,” describes the surprising difficulty of reestablishing prairie ecologies. Chapter 5, “Coevolution,” traces the circular nature of cological networks of dependency and evolution. Chapter 6, “The Natural Flux,” describes how living systems maintain stability by riding the crest of a wave of instability. Chapter7, “Emergence of Control,” relates these notions of closed loops and constant change to the cybernetic concept of control in artificial systems. Chapter 8, “Closed Systems,” returns to biological ecology by examining instances of sealed self-sufficient biological systems, and chapter 9, “Pop Goes the Biosphere,” reports on the Biosphere 2 experiment on closed ecologies that include humans.
The next four chapters apply these ideas to developments in artificial systems. Chapter 10, “Industrial Ecology,” envisions the impact of artificial life on communities of machines such as offices and factories. Chapter 11, “Network Economics,” discusses the impact of cheap, long-distance, high-bandwidth communications on such communities. Chapter12, “E-money,” describes encryption and related technologies that will permit networks to support the financial aspects of business. Chapter 13, “God Games,” shows how simulations can help people manage large complex systems whose behavior is emergent and thus not fully known in advance.
Thus far, the book concentrates on the interactions of entities in a living system, and how artificial systems can emulate these interactions. The next six chapters discuss how living systems, and artificial ones that imitate them, evolve through time. Chapter 14, “In the Library of Form,” motivates evolution as search through a complex space, using Jorge Luis Borges’s metaphor of a vast library containing every book that could be written. Chapter 15, “Artificial Evolution,” reviews several milestone experiments in computational evolution. Chapter 16, “The Future of Control,” discusses applications of emergent behavior to video animation. Chapter17, “An Open Universe,” develops the notion that evolution not only permits one to search large spaces, but also opens up entirely new spaces to search. Chapter 18, “The Structure of Organized Change,” shows how Lamarck’s once-repudiated notion of the inheritance of acquired characteristics is becoming an important model for evolutionary change. Chapter 19, “Postdarwinism,” squarely faces the failure of evolutionary improvement of individuals to explain the development of widely differing types of organisms, and speculates about other mechanisms that might be invoked.
A background theme throughout most of the book is the pervasiveness of complexity and instability in both natural and artificial life. The next three chapters present a counterpoint to this theme, showing various forms of stability and order within complex systems. Chapter 20, “The Butterfly Sleeps,” is devoted to Stuart Kauffmann’s autocatalytic networks. Chapter 21, “Rising Flow,” generalizes beyond Kauffmann’s results to posit an overall purpose and direction of evolution toward greater evolvability. Chapter 22, “Prediction Machinery,” discusses how the existence of local short-term order in chaotic systems may permit prediction in domains as concrete as the stock market.
The final two chapters are a retrospect and a prospect. Chapter 23, “Wholes, Holes, and Spaces,” traces the history of the ideas brought together in the book back to the cyberneticists. Kelly discusses why cybernetics did not deliver on its early promise, and outlines a series of questions that remain in spite of the recent explosion in complexity studies. Chapter 24, “The Nine Laws of God,” abstracts nine principles that recur throughout the book, and that form a rudimentary guide to engineering systems with lifelike characteristics.
To an engineer, this amorphous tome is often frustrating, but I suspect that is a large part of its value. If we are to learn to grow systems instead of simply building them, we must become comfortable with their densely tangled nature. Kelly’s potent brew of computer science, biology, and theology may serve as the first engineering manual for neo-biological civilization.
