Level-headed work

At 43, Daniel Hillis is one of the genuine stars of the contemporary computer industry. As an undergraduate at the Massachusetts Institute of Technology in the 1970s, he worked at the Logo Laboratory developing computer hardware and software for children. He also designed computer-oriented toys and games for the Milton Bradley Company. As a graduate student at the MIT Artificial Intelligence Laboratory, Hillis designed tendon-controlled robot arms, a touch-sensitive robot "skin" and a computer composed entirely of Tinkertoy, which is now on display at the Boston Computer Museum.

Hillis soon began to study the physical limitations of computation and the possibility of building highly parallel computers. His work culminated in 1985 with the design of a massively parallel computer with 64,000 processors, the topic of his PhD thesis. He called it the Connection Machine.

Hillis co-founded Thinking Machines Corporation to produce and market the Connection Machine. His interest in building gadgets and games was influenced by his friend, the Nobel prize-winning physicist Richard Feynman, who would leave Caltech in the summer to go to Cambridge to work for Hillis at Thinking Machines. Thinking Machines folded in the early 1990s and morphed into several companies, one of which does data mining and was bought by Oracle.

Meanwhile, Hillis was making the journey from hard-nosed hardware design into the world of computer-assisted entertainment. In 1996 he became the first Disney fellow. He now has the very un-Mickey Mouse job of vice-president of research and development at the Walt Disney Company.

The Pattern on the Stone is a readable introduction to the basic, familiar ideas of computer science, and to some new ones as well, such as quantum computing and the evolution of computers. Three themes drive the book. First, the idea that we can, and perhaps must, design computers hierarchically, dealing with problems at one level without worrying about what happens at other levels. To a large extent, you can design circuitry without worrying much about the software that will run on it, and vice versa.

This is the result of a second theme, Alan Turing and Alonzo Church's 60-year-old almost universally believed thesis that there is really one kind of computer, the universal Turing machine, whose essential behaviour is independent of the materials we use to construct it.

Finally, there is a third theme, a partial denial of the first, that there may be a way of evolving computers rather than designing them, and that maybe this is the only way to build really complex computers that might have minds.

So why do I find the book less than entirely satisfactory? Because it should be as brilliant as Hillis could make it, and taken as a whole it just is not. The early chapters on Boolean logic and algorithms are standard material, but the examples and illustrations are not particularly revealing or easy to follow. At this point there is also some careless editing. The logical symbol for "or" is written as the words "down caret".

The latter parts of the book are where Hillis has the chance to speculate,on quantum computing, heuristics, parallel computers and evolutionary computing. It is these chapters that redeem the book and make it worth putting in the hands of a student who already knows the basics of computer engineering and wants a taste of what the next century may bring to information technology. The Pattern on the Stone is a good book, but one that should have been even better.

Peter Gibbins is executive director, Digital VCE, and visiting professor, University of Exeter.