Now we could turn our attention to the esthetic aspects of the Connection Machine: how did we view it, what did it mean to us, what did we want to say about it? We wanted a strong, simple form that had meaning, that expressed the essence of the machine -- we wanted to let the machine speak for itself.

Our goal seemed to be clear: we wanted to show how the machine worked. The most obvious solution seemed to be to expose the interior of the computer, baring the boards and cables that made up the machine. But every computer had boards and cables -- how could we show that this machine was quite different?

Just at this time Sidney Lawrence published an article called "Clean Machines at the Modern" in the magazine Art in America. Lawrence's article described the evolution of the idea of functionalism, showing how the Museum of Modern Art in New York had elevated the Bauhaus emphasis on "material and proportion rather than applied ornament" (6) to serve as the exclusive canon of good design and good taste. In MoMA's dictum I recognized the source of our own ideas on design: "it was precisely the inner workings of mechanical objects that offered an appropriate standard and inspiration for contemporary design.... Functional design should expose and clarify function, not disguise it." (7)

Modern architects used this paradigm to lay bare the basic components of building: space, load-bearing members and building materials. Product designers dealing with exclusively mechanical products, where the physical form largely determines the function, could also implement this principle with ease. A mechanical device expresses its function in the visible world, it is graspable, and moves or physically affects its operators and their world. We, however, were working with a very different sort of product.

Our dilemma was succinctly described in the same article: "[electronic machines] are incomprehensible unless one knows about the existence of invisible forces ... [they] do not visually explain themselves." (8) Indeed, both a simple text processor and a powerful supercomputer are composed of exactly the same elements -- chips, printed circuit boards, and cables -- and everyone except an electronics specialist will see a difference in quantity, but not quality, between these two extremes. As Lawrence concluded, "The jumbled appearances of a computer circuit, in fact, tell us nothing at all about its function." (9)

The MoMA design department has addressed itself to the esthetics of electronic components and has even exhibited the schematic beauty of integrated circuits and printed circuit boards. This fascination with the internal components of electronic devices gives no help, however, to the designer confronted with the task of developing the external appearance of a machine. Still, MoMA's design department gave us an important starting point with its characterization of the esthetics of electronics as "the dematerialization of finite shapes into diagrammatic relationships." (10)

It became clear to me that we had to extend the meaning of "function" beyond physical structure, beyond the purely mechanical into the abstract. Indeed, for people who work with computers, the image of the machine in their minds has nothing to do with boards and cables. Instead, they see the conceptual structure of the system, the "diagrammatic relationships" which can vary in function and detail in the same way that human habitation can vary from a rough, simple shelter to an ornate and complex palace. To truly understand the function of a computer one has to look at the schematic representations computer scientists use to talk about the architecture of a computer, the structure buried in microsopic layers of silicon and hidden in mazes of electronic circuitry.

This extension of the definition of "function" was our first departure from the tenet that mechanical machines should provide "a standard of reference for judging contemporary design." (11) The second departure was a rejection of the sterile, cold utilitarianism that the term "functionalism" has come to represent. We didn't want to build a computer that looked just like a refrigerator or a washing machine, even if that was the most "practical" and "functional" way to package it. We wanted the design to express the excitement we felt about the machine and about its potential to revolutionize computer architecture.

For many people, computers are bloodless beige boxes with incomprehensible electronic displays that never work anyway, especially when one is in a hurry. Scientists involved in computer research, on the other hand, see computers as the tools for building new worlds; they see themselves as pioneers and settlers in a wilderness conquered not by the plow and the rifle, nor by space ships and laser guns, but by mathematics and programs, by the brain and by abstract thought. In this community people often refer to acquaintances by their electronic mail addresses, and many friendships -- and even marriages -- have started on the world-wide communication network that instantly connects research labs in different cities into one great electronic watering hole.

We at Thinking Machines were all members of this electronic tribe. For us the "electronic village" eagerly awaited by preachers and gurus of the Information Age has long been a reality of daily life. The activities of this "daily life" are a semiotician's dream -- or nightmare: the members of these tribes spend their time creating signs with no physical referents, systems of signs that mean exactly what their inventors wish them to mean, and worlds that function according to rules made up by their creators.

Artists and natural scientists also create intricate systems of signs or symbols, building worlds of their own that an outsider must study in order to understand. Why do lay people consider artists and natural scientists interesting, if slightly weird, while they view computer scientists as simply wierd? The purpose of both art and science is to develop and communicate insights into the physical and spiritual world we all share. The systems of signs that computer scientists build, however, are actually schematic descriptions of the insides of machines and, as such, are self- referential rather than interpretive or representational. The double-helix model of the structure of a human gene, for instance, describes something that is part of each of us, but a diagram of the message-passing network of a computer describes something that is part of a machine in which only a specialist has an interest. In the design of the Connection Machine, we wanted to express the mystery of the world of computers in a way that would capture the imagination of all who saw it.