According to the US National Highway Transportation Safety Administration, airbags have saved 1,500 lives. An unintended consequence is that airbags have killed 22 infants. Airbag systems are being redesigned to avoid this, but the redesign itself may have further consequences. Very likely they will be fewer than the 22 deaths in the same proportion as the 22 are fewer than the 1,500. It is rare good fortune when the numbers to make comparisons are available.

Tenner's book is about the unintended consequences of technology- which have turned out so far to be less significant than those of many political and social decisions, such as Neville Chamberlain's "peace for our time". Unlike most writers on this subject, Tenner is judicious, often concluding that some specific innovation was worthwhile in spite of some unfortunate consequences.

Tenner's research is more thorough than that of his predecessors. In some cases, quantitative comparisons are possible, as in the airbag example, and it is a pity that he did not do more of them.

The lessons to be learned from his topics are various. Here are a few. The gypsy moth was brought to the US in the 19th century in the hopes of founding a silk industry. There is no industry, and the moth's damage to forests requires expensive control measures. The moth was a disaster, but was it a mistake? It was one of a number of introductions of foreign plants and animals, many of which have been extremely successful. Was there knowledge at the time that should have told Leopold Trouvelot not to do it?

É«ºÐÖ±²¥

The fire ant was not an intentional introduction but a side effect of transportation from South America. The efforts to control it have not been very successful and may have been counter-productive by killing some of its enemies. My own technological optimism suggests that a way of eradicating the gypsy moth and the fire ant will be found eventually.

Eucalyptus trees were imported from Australia into many countries. It was thought that they would make good timber trees, but the wood from eucalyptus grown in the US splits too easily, apparently because the trees grow too fast. They provide good windbreaks and are still widely planted in California, where they are the tallest fast-growing tree in many parts of the state. They are controversial, but many people argue from the principle that no nonnative can be any good.

É«ºÐÖ±²¥

Most of us are not so squeamish, and the world would be a poorer place if the kings of Spain and Britain had taken advice from the future to forbid bringing plants or animals from one continent to another. There would have been no horses or cattle for the cowboys and Indians, and no potatoes for the Irish.

Antibiotics were introduced in the 1940s, and bacteria resistant to their effects immediately began to evolve, necessitating ever more new (and more expensive) antibiotics. Tenner cannot resist a bit of scapegoating here. So far we are ahead of the game. Tuberculosis once killed 300,000 Americans per year. After declining to a very low level, it is now back up to almost 2,000 - mostly Aids sufferers and other immune-deficient people.

Tenner's most interesting complaint concerns computers in offices. First, he considers that productivity has not increased. I do not know whether the following counts as a productivity increase, but I find it much less frustrating to deal with an organisation where the person who answers the telephone can call up your record.

Second, the amount of paper has increased, belying the idea of a paperless office. It would have been surprising if the advent of cheap, high-quality laser printers had not led to more printing. Now we all have Winston Churchill's extravagant luxury of making our corrections on page proofs.

É«ºÐÖ±²¥

Tenner tells us that people print their email. I did some polling at Stanford and at IBM. Engineers and scientists almost never print email and store it permanently on the computer. Secretaries, administrators and executives often print it, and a few even file it in printed form. Mostly they have good reasons, but advances in notebook computers will obviate them. We shall have the paperless office yet, and we shall be able to read from screens lighter than a book in bed or in the bathtub. Surely someone will match the old advertisement that ballpoint pens let you write underwater with a new one asserting that the new screens let you read underwater.

Tenner's most interesting complaint is that personal computing environments have become so complex that ordinary users need the help of experts to survive, and that people with important jobs often spend a lot of time helping others use the computers - time that would be better spent in doing their jobs. He is right about that. I am a computer scientist and have been using computers for 40 years. Nevertheless, I have had to rely on my graduate students and our computer facilities professionals to deal with the complexities of configuring collections of programs. The problem has got worse in recent years with the advent of graphic user interfaces and programs that interact with others over networks; and there are trends that may make it worse. Feasible technical fixes exist, but they may also require some standardisation of how computer programs are configured and used. The companies that adopt them will get big competitive advantages.

Tenner's main recommendation is eternal vigilance - a solution that was also recommended in the political sphere.

John McCarthy is professor of computer science, Stanford University.

É«ºÐÖ±²¥