Nature lessons for computers

Computers are everywhere: inside cars, washing machines, in offices, entertainment, hospitals, finance and bureaucracy. Who would have thought that the research that led to the worldwide web would be commonplace in ten years? But although computers pervade every activity, we find them difficult to use and unreliable; one need only mention the daily problems with email spam or crashing computers to see that there is a long way to go.

While society is still coming to terms with computers, they are about to change again and become even more entwined with human life.

Computers can imitate biology and biology computers. Cyborgs are just one idea from science fiction that could become reality; what about biocomputing viruses for medicine? To survive, life has continually to solve many problems. Evolution solved the problem of finding successful life forms; immune systems solve the problem of microscopic attack; brains solve the problems of where the next meal is coming from; eyes solve the problem of what to chase; and herds or swarms solve problems of social survival.

All these biological ideas are fundamentally computational. They can inspire new ways of harnessing computers to solve problems by simulating the approaches life takes. Indeed, some very successful computer "tricks" have been developed from artificial neurons. And it works the other way around: we do not need to use traditional silicon and digital circuits to make computers work, we can take ideas from biology and use DNA or other chemicals to do the work.

Nancy Forbes's book Imitation of Life is an overview of the wide-ranging activity in these exciting areas. Ten key chapters cover the spectrum of research: neural networks, evolutionary algorithms, cellular automata, artificial life, DNA computation, biomolecular self-assembly, immune systems, biological hardware and biology as computing. Many of these areas were first charted by the historical giants Alan Turing and John von Neumann in the 1940s. Their ideas are just starting to become a reality.

Turing was British and von Neumann was American, having emigrated from Hungary, and it is striking how much of the research Forbes reports is international. Sadly, international is now more a euphemism for not done in the UK. One conclusion is that we in the UK (and indeed everywhere outside the US) should start teaching this material to undergraduates before we get left behind. Perhaps the current generation of undergraduates can get involved? Imitation of Life would make an excellent seminar book for first-year students to work through, perhaps to complement a more practical book such as Gary Flake's The Computational Beauty of Nature (1998).

Indeed, the style of Imitation of Life lends itself to seminar study - I think otherwise it falls between the two high stools of popular science and textbook exposition because it is at once too technical to be popular yet has too little working detail for self-study. For example, we are told that Georgia Tech computer scientist Dick Lipton has an efficient method for solving a problem called SAT. Although the journalistic style of being told where Lipton works sounds helpful, we are not told any details, so we do not learn anything useful. It does not provide references either. But it is enough to search the web yourself - I looked up Lipton on Google and found that he is doing work on email spam. So in seminars the bird's-eye approach of the book can be turned into a stimulating way of learning more, motivating learning about cutting-edge research.

Imitation of Life ends with an all-too-brief four-page epilogue discussing the impact of the life-changing developments the book has reviewed. We desperately need ways to make computers more reliable and more useful. We will also spin off many innovations in biotechnology, medicine, nanotechnology, and there will even be escalation in cyberattacks and biodefence systems. In short, the imitation of life will have an unknown impact on life itself, on humans and our society. I hope, then, that this short book helps transform undergraduate curricula in computer science departments.

Harold Thimbleby is professor of computer science, Swansea University.