We have a lot to learn from ants

As anyone who is familiar with computer games will know, computers can create universes that simulate reality to build a world in which to play. Given instructions, they can simulate alien attacks, flying aeroplanes or busy life in a city. When the rules of computer universes are based on physics, we see roughly what we expect. When, however, the rules are based on biology, we get something surprising. Things look alive. We get digital biology.

Just as evolution helps nature find solutions to surviving effectively, computer simulation of evolution allows us to find solutions to survival in imaginary worlds, which we can design to represent useful problems. Just as one might imagine a bacterium evolved to suit some special habitat, we can write programs that evolve to fit financial habitats, finding out as they do so how best to work the markets.

Peter Bentley is well known for his work in evolutionary computing, applying insights from biology to computer-based applications. The gist of Digital Biology , which is his first popular-science book, is that biology follows rules and that computers can be programmed to follow the same sorts of rules. By copying biological rules, we can harness "nature's creativity" to solve our problems, provided they are suitably expressed in universes created inside computers.

Imagine an ants' nest. The ants as individuals may not be very clever, but by working together they forage and find optimal routes to supplies of food. Likewise, a computer can be programmed to simulate what ants do and can then solve hard problems such as finding the best way to visit customers to keep a company in business. Similar nest-inspired digital biology methods have been used to schedule processes, to route automatic vehicles and to design electronic circuits.

It is hard to write popular science, especially when combining two such contrasting subjects. Some will find the explanations in their own field simplistic, yet will not get enough from the book to go deeper into the other field. The writing has a lot of imaginative story-telling, but it is not obvious how to get from inspiration to involvement. You can admire the creativity of researchers and be fascinated by the applications, but it is hard to see how to do any digital biology for yourself.

Fortunately, there is plenty of digital biology to explore on the web, although the book does not point this out. Apart from a few asides, the book also fails to explore vision or robotics, which are the main ways in which computers sense and interact with the real world.

Perhaps because his digital biology is inside computers, Bentley is keen to make sure we realise that it is supposed to be real biology rather than mere simulation. I happen to disagree, but there is no doubt that harnessing digital biology to real-world problems, being inspired by evolution, neurons, insect swarms, immune systems and embryology is going to change our lives - and probably more dramatically than robots will.

Harold Thimbleby is director, University College London Interaction Centre.