The physical scale in which information is processed in computers is enormous. Millions of logic gates, the building blocks of all the complex tasks performed by digital machines, are present in the area the size of your fingernail. Without the extensive miniaturization involved, desktop computers would occupy whole buildings. Miniaturization in the conventional sense, of making the same tools smaller and smaller, is running into a kind of speed limit. The speed of light limits the speed any electric impulse may travel a certain distance. One way to get around this limit is to bring the components closer to each other. Unfortunately, the logic gates of silicon, though sometimes replaced by smaller and faster ones of other semiconductors such as gallium arsenide, are already packed as tight as sardines. Some particularly futuristic scientists think that a change of thinking is required. Instead of making components out of large quantities of material such as silicon, they could be made from scratch, atom by atom, a billionth of a metre in size.

Nanotechnology is the science, art, and dream of reducing the scale of machines to a fraction of their size, on the scale of a nanometre -- a billionth of a metre, and a 100 times the length of an average virus. This can be done using advanced molecular simulation and design tools to actually build, for example, a motor, an atom at a time. Together with detailed programming through on-molecule computers, swarms of machines the size of specks of dust could form floating factories the size of a matchbox. Rather than attempting, as is common at present, to create and finish tiny components in huge monsters of steel and concrete, these nano-factories could build from the molecular level, assembling cars apparently out of thin air.

Nano-robots could be used for just about anything. Living at a molecular scale, these would easily build advanced materials, microprocessors, and other nano-robots without the problems we might have with manipulating individual atoms. Their on-board computers, programmable in a way DNA can never be, provide us the flexibility to use the same robots for building drugs, cleaning up pollution, or traveling along the bloodstream to repair tissues and kill cancer cells. The promise of nanotechnology is fantastic, whether viewed as a paradise or nightmare.

Ever since Eric Drexler at Xerox PARC outlined the concept in his book, The Engines of Creation, seven years ago, nanotech has been both applauded as the next technological revolution, and criticized as absurd and impractical. Currently, it is extremely difficult to manipulate molecules, leave alone atoms, in the contrived ways needed to build machines out of them. In 1990, Don Eigler adapted the scanning-tunneling electron microscope (STM) to build a molecule forming the IBM logo. Since then, researchers have tried various techniques to grasp and move atoms, but it's a long way before anything significant will happen.

The first spin-off products of nanotech research are likely to be in information technology. Molecular devices could pack far more components into the same chip, and memory storage could go up by a factor of a million. Molecular-sized sensors could be used to provide special information about chemicals and compounds, for instance, in medicine.

Making a molecular system with moving parts is likely to come much later. Moving parts are absolutely essential to any manufacturing or fabrication process, but their unwieldiness is a reason for them lagging behind information processing. Funnily enough, it is with moving parts that some of the only existing nanotech machines exist -- the microscopic flagellum of a bacterium acts just like an outboard motor, propelling it forward, controlled by its own on-board processor, the DNA.