A Laureate’s Lament

The late Herb Simon, winner of the 1978 Nobel Prize in Economics for his pioneering work in organization theory and long-time Professor of Computer Science and Psychology at Carnegie Mellon University, was a man at home in many different temples.

His concern was that the natural sciences had edged the artificial sci­ences out of their rightful role in professional education—in engineering, architecture, business, education, law, and medicine—to the point that the very term “artificial science” sounds odd. Though the artifacts of artificial science shape our everyday lives and social interactions, pro­fessional educators weren’t giving them the respect that they deserved. Fortunately, Simon’s expertise in conquering organizational challenges also led him to detect a solution stirring.

A science of artificial phenomena is always in imminent danger of dissolving and vanishing. The peculiar properties of the [artificial] artifact lie on the thin inter­face between the natural laws within it and the natural laws without. What can we say about it?... The artificial world is centered precisely between the inner and outer environments; it is concerned with attaining goals by adapting the former to the latter. The proper study of those who are concerned with the arti­ficial is the way in which the adaptation of means to environments is brought about—and central to that is the process of design itself. The professional schools will reassume their professional responsibilities just to the degree that they can discover a science of design, a body of intellectually tough analytic, partly formalizable, partly empirical, teachable doctrine about the design process...

Decades of experience have proven Simon half right; his diagnosis was correct, his prognosis flawed. The artifacts of artificial science remain central to our lives as social creatures. At least two of them—markets and computers—have emerged as central features of everyday life. In addition to everything else that they’ve done for the modern world, they combined to give us the information sector. But academic computer science programs hardly developed in the direction that Simon predicted. The luminaries of academic computer science chose to burnish the engi­neering credentials implied by “computing” and to abdicate the social responsibilities implied by “information.”³ Both the teaching and research emphases of most computer science departments (typically located in schools of engineering) stress the construction of faster, slicker machines and the software needed to make them soar. Social impact issues, studies of the ways that people use computers, or inquiries gov­erning the artificial interface between the natural sciences of silicon-based computing machines and carbon-based computing users, have been pushed to (and often beyond) the periphery of the field.

The information sector lies precisely at that interface. It exists as an interface artifact of artificial science. While this interface artifact assumes many faces and goes by many names, four of its most famous personae played critical roles in the information sector’s formative years: Windows, Linux, Napster, and the World Wide Web. Windows and Linux are interface artifacts that separate human users from their own personal computers. Napster is an interface artifact that sits on a com­puter network and separates two human users from each other. And the World Wide Web is an interface artifact that separates human users from the vast resources of the Internet.

Software is an internal computing matter. Software is a set of instruc­tions that guide electrical impulses inside a computer. The natural sci­ences used to study computers answer the questions about software’s identity. Computer scientists study the phenomena of software and com­puters; they’ve derived a complex and elegant science that is of direct interest primarily to those who choose careers in computing. The public is generally content viewing their outputs as black boxes without under­standing their inner workings.

How software developers are motivated and rewarded is a human matter. Computer scientists rarely trouble themselves with such issues; they prefer to leave these concerns to their colleagues in economics, psy­chology, management, or business. These issues are essentially matters of human cognitive psychology—appropriately described by the natural 4 sciences.

But how software evolves—therein lies the interface question. Evolu­tion itself, though perhaps the quintessential internal response to the changing needs of survival, is inherently an interface phenomenon. All organisms exist within an environment with which they must interact. Sometimes the environment changes. When that occurs, the organism must either adapt or die. At other times, the organism changes first. It gains new abilities to interact with its environment. When it applies those abilities, the environment may change, too.

Two organisms and two environments define the information sector. In a stunning bit of symmetry, each organism defines the environment in which the other must survive. The first organism is human; the com­puter’s requirements define its environment. That selfsame computer is the second organism; it must exist in an environment comprehensible to the human. Failure by either party will lead to mutual irrelevance; neither can survive in the information sector alone. A human unable to interact with the computing environment loses the benefits of modern technol­ogy, while a computer unable to interact with the human environment is a worthless box. Their existence as functioning, productive organisms of the information sector is entirely symbiotic.

Software defines the interface between these symbiotes. It must adapt and evolve in response to their needs. Changes in human aptitudes or even tastes may force one sort of evolution. Changes in computer capa­bilities may lead to a second. And whichever organism changes first may—through the evolution of the software at the interface—impel the other into its own evolutionary change, in a potentially perpetual cycle of adaptation and growth.

The only way to understand software evolution is to explore the nature of software itself. First and foremost, all software derives from a single goal: the desire to increase the range of activities with which a computer can prove to be helpful. This goal leads to a view of software develop­ment as translation. Computer scientists face a daunting task. A com­puter is a machine capable of determining when voltage levels rise and fall. A potential computer user is a person who expresses himself or herself in an imprecise human language (usually English, or increasingly, Chinese). An actual computer user also believes that the computer will help solve some task.

Anyone with even a cursory exposure to computers realizes that there are actually two series of translations in this chain. Responsibility for the first half lies with the human user, who must learn some combination of a restricted lexicon (e.g., a programming language), Boolean logic (i.e., stringing keywords together using “AND”s and “OR”s), and graphical manipulation (e.g., mouse clicks) to talk to the machine. Responsibility for the second half lies with engineers and programmers. Their work begins by representing high and low voltage levels as 1s and 0s, respec­tively. They then group these binary digits (“bits”) together to generate more interesting numbers that encode words. Of course, only a small subset of these words is meaningful to the computer—those that it can use to manipulate yet other voltage levels. This limited lexicon and grammar form the basis of a low-level computer language (e.g., an

assembly language), with which a talented programming language designer can develop a “higher level” language.

While people always needed a full translation chain to use computers, the locus of technical and commercial attention has shifted as comput­ers have matured. In the 1960s, virtually all computer users were tech­nically trained professionals personally proficient in Fortran, COBOL, ALGOL, LISP, or some other specialized language that bore only a cos­metic relationship to English. By the 1990s, these once-popular computer languages had given way to more sophisticated, object-oriented lan­guages like C++ and Java—which still bore only a cosmetic relationship to English. Of greater significance, though, advances between the 1960s and 1990s enabled humans with no understanding of these odd-looking dialects to communicate with computers.

The bridge connecting the human and computer organisms is thus con­structed of one translation chain built “up” from voltage levels into increasingly complex languages that look more and more like English, and one translation chain built “down” from English to look more and more like math. These chains meet at a level known as the “user inter­face,” literally the point at which the user interfaces with the machine, but also the artificial interface artifact that Simon urged us to study.