Under The Microscope

OUT OF CONTROL: THE NEW BIOLOGY OF MACHINES by Kevin Kelly; Fourth Estate, London, 1995; 666 pp; $25.00 (paperback)

Reviewed by Craig Webster

This book is a fascinating roller-coaster ride through a host of emerging technologies which will no doubt have an influence on all our futures. Kevin Kelly, executive editor of Wired magazine and "Internet guru", demonstrates quite convincingly how the technological is becoming more biological.

Artificial intelligence, robotics and our knowledge of ants and bees has produced insect-like robots capable of smart collective behaviour. Genetics, evolutionary theory and massively parallel connectionist machines (the fastest computers on the planet) are yielding emerging fields like evolutionary software design where the computer code is "bred" rather than being written. Open, closed, complex, self-organising, centrally controlled and distributed systems are all examined and contrasted, including everything from Borgian libraries to zero-sum games.

Kelly contrasts the paradigmatic differences between the made and the born. What is made by us tends to be minimal, mechanical, predictable and maintenance intensive (even in our "autonomous" systems). By contrast, when we consider the different magnitudes of information in a blueprint compared with a DNA strand, we see that the born is vastly more complex, organic, unpredictable and constantly adapting to environmental changes.

Many fascinating comparisons are made between the organic and the computational. What guides the behaviour of living things? Are species computing optimisation problems within their neck of the genome?

Maybe so, but since evolution concerns itself only with end results and not with elegance of method, the way individual species go about solving these optimisation problems is very different to the way we perform them with mathematical theorems. Evolution employs incredibly cluttered, "noisy" methodology while we always attempt to find the simplest solution. Einstein's adage of "as simple as possible, but not simpler" just does not apply to evolution, which is guided only by the principle of "if it works, stick with it". Evolutionary problem solving, Kelly claims, holds the key to solving many problems which are not addressable in minimal, simple terms. However, the evolved solutions to these problems, while they may work, will more than likely be stunningly incomprehensible.

We are left in the rather uncomfortable position of employing an automated method which yields a solution to a problem that we cannot solve, and does so in terms that we do not understand. Kelly believes this is how the new biological machines will get out of our control. We will quickly not be able to follow what they are up to, but may also become more and more reliant on them to yield solutions in areas conventionally considered insoluble or intractable.

Kelly is certainly optimistic about the future that these new information technologies make possible, but also presents some of the consequences of this new synthesis between our machines and biology. The book on the whole is accessible and a real technological page turner. It will be of particular interest to anyone with some background in computing, artificial intelligence, biology, information theory or cognitive science.

Craig Webster is currently a clinical researcher in the Anaesthetics Department at Auckland's Green Lane Hospital.