Is computer science a science? What is at stake in the question of “the sciences of the artificial”? Herbert A. Simon was a distinguished professor for five decades at Carnegie Mellon University, one of America’s most elite and important technology institutes of higher education. Simon won the Nobel Prize in economics and the Turing Award, which is the most prestigious citation for achievement in computer science, given annually by the Association for Computing Machinery. Simon’s work was highly interdisciplinary, ranging from economics and psychology to Artificial Intelligence and the study of large organizations and complex systems. In his book The Sciences of the Artificial (a pioneering work first published in 1969, then subsequently revised in 1981 and 1996), Simon does not seek to formulate any fundamental philosophical definition of the array of sciences which study – to invoke his terms – the “man-made” as opposed to what is “given by nature.” He searches for ways to pragmatically identify the characteristics of the artificial sciences. His approach is imbued by American pragmatism.
What makes some phenomena and systems artificial, according to Simon, is not only or primarily the fact that they come to be through artifice, design, engineering, or other human cultural practices, but rather the pragmatic and operational circumstance that they interact with their environment. This quality of the objects studied by the sciences of the artificial that they continually engage with their environment renders them more dynamic and changeable over time than the phenomena studied by the natural sciences.
It is more difficult to make statements that remain valid for a long duration of time about synthetic entities. The contingency and malleability of artificial phenomena, according to Simon, are due to their deep and continuous involvement with their environment, in contrast to the “necessity” which is a chief property of natural phenomena. The latter exist in a subordinate relationship to the power of natural laws. This difference has unfortunately often led scholars and thinkers to mistakenly regard artificial materials and systems as not “falling properly within the compass of science.” The challenge that Simon defines is to pinpoint exactly how can one make valid empirical propositions about things and systems which behave varyingly in their permanently changing circumstances and whose behavior is different if observed at different times.
According to Simon, the thorniness of the problem of artificiality – which affects many disciplines extending from economics, management, and information processing to education, engineering, and the cognitive psychology understanding of thinking and problem solving – is also due to the normative character of the objects inquired into by these fields. These sciences are concerned not only with “how things are but with how they might be.” There are ethical, political, economic, and purposive-rational goals involved in the investigated occurrences. Thus, Simon elevates design to a central position in his framework. The mission of creating a science of artificial is inseparable from the task of creating a science of design. Design is the key to grasping and intervening into how systems abide in complex environments.
In her 2009 book Simulation and Its Discontents, MIT professor of the social study of science and technology Sherry Turkle reflects on the transformations in scientific, engineering, and design education at MIT that occurred when computers and software were introduced to all fields of study in the 1980s and 1990s as major and intensive components of the learning curriculum. Turkle concludes with regret that many thinking skills and significant knowledge were lost in the training of scientists and professionals when all disciplines came to increasingly resemble each other in their shared emphases on simulation and visualization. The way of working – without computers – of older professors who were retiring was more direct and less mediated. Herbert A. Simon, contrary to Turkle, sees the computer as being a fantastic development for its stimulation of interdisciplinarity. He praises “the growing communication among intellectual disciplines that takes places around the computer… All who use computers in complex ways are using computers to design or to participate in the process of design.” The computer becomes the tool par excellence for transdisciplinary design. In Simon’s vision of the inter- or trans-disciplinary, there is no place for the abiding value of the mono-disciplines.
What is important about artificial systems for Simon is their goals, functionality, self-organization, normativity, capability to adapt to new circumstances, and orientation towards how things should be. The artefact is performative in its interaction with its environment. Simon sums up his position with the concept of interface. There is an interface or meeting point between the inner organization and the outer environment of the artificial entity which underlies its design or intended purpose. Simon claims that his conceptual framework has the benefit of being predictive. Insights into the goals and behavior of the artifice enable an anticipatory advantage in foreseeing the future. He cites the state of homeostasis in biology (the steady internal conditions maintained by living systems) as an example to support his thesis.
Simon recommends positing an invariant relationship between inside and outside for heuristic purposes, isolating the inner system from its outer environment in a temporary bracketed way. The interface between inside and outside should be designed simply and elegantly, therefore strengthening its qualities of powerful abstraction and general applicability. The priority of interface then leads to the potency of simulation. Once we have a clear grasp of the interface, then that interface can be simulated “as a technique for achieving understanding and predicting the behavior of systems.” The computer, with its ability to simulate, model, and try things over and over, becomes the ideal tool for the sciences of the artificial.
Simulation, according to Simon, can provide amazing amounts of new knowledge. Software can imitate human behavior in many domains, given an algorithmic description of the behavior. “No artifact devised by man is so convenient for this kind of functional description as a digital computer.” It is the ideal device for the empirical social sciences, the exploring of the consequences of alternative independent variable values and organizational assumptions. The design of software, according to Simon, is a behavioral process. The software program is a logical arrangement of symbols to be manipulated by a program-control component (a Turing machine). In any design or conceptual phase of the software development cycle, not much can be known about how the software is going to behave. You build the software and then you see later how it behaves. Things become known in doing and trying out, in observing what happens when the software is up and running and interacting with its environment.
Herbert A. Simon provides an empirical methodology for the unification of the social and informational sciences. Simon wants predictability of the entire artificial world-ambience. His argument is a visceral and inaugural rejection of any philosophical approach. He is interested in the acquisition of useful knowledge for the scientific management of that simulation model that we call society. Yet his highly influential approach dissuades us from asking the crucial question: how can the philosophy of science be applied to computer science?
Simon presciently asked the crucial question “What is at stake in the question of ‘the sciences of the artificial’ as separate from scientific approaches to the natural world”? However, his strictly empirical methodology led him away from all philosophy, and from any engagement with the philosophy of science to help in answering the question.