1.3 MODELS FOR ENGINEERING DESIGN
With the appropriate processing environment about it, text can be regarded as an interactive, social medium. Whether this interaction be between the internal musings of the individual or the unorganized debate of a crowd, language and text provide the flesh and form. The same is true of design. As hypertext, hypermedia, virtual reality, synthesized design, advising agents, and remarkably realistic visualization engines enter as elements, the computational paradigm for computers still applies. But add multi-media, fax technology, video-conferencing, and other high-bandwidth information, leaving interpretation largely between humans and not computers, and the results go beyond that of a mere computerization of text or of design.
Rather than simply assisting the way humans think, computers can now model or assist in the way humans communicate. The results are what Edward Barrett describes, in direct contrast to the computerization of text under the computation paradigm, as "the textualization of the computer": the computer is used primarily as a communication medium rather than a tool [Barret 89A]. The computer environment and its signs and symbols become, at the very least, as effective a means of the formation and communication of ideas and meaning among a culture of users as is language and text alone. This in turn brings forth the concept of shared memory: a common repository of the past, present, and future through which multiple cultures can communicate.
Engineering design and computer models in general can be thought of as a common text or a type of shared memory. Alvin Toffler noted the need for and use of complex scientific computer models, but also saw that the inner workings of these models were removed from direct inspection by the non-scientist. The result is that unchallenged assumptions can be embedded into the model, and these assumptions may fundamentally affect the model's results. As such, models can be used toward political ends not endorsed by reality [Toffler 90]. More importantly, political and scientific debate is often diverted to center around the accuracy of the model, rather than on insights the model may produce. Some formalisms must be built into these models throughout all levels to permit every crevice and corner of assumptions and equation sets be exposed fully to light.
"Critical theory" addresses the development of models, particularly within literature and the domain of public policy. Apart from attempts to cast them as sciences, engineering design and public policy have many of the same traits. Both must simultaneously resolve diverse, and sometimes conflicting, criteria. Both rely heavily on models which attempt to depict the real world, as in the statistical forecasting of a disease epidemic or the numeric simulation of molten plastic flow. Both involve decision-making processes (ideally cognizant of the degree of uncertainty in the models used), because there is never just one optimally "correct" design or policy, but rather a range of satisfying options.
Critical theory recognizes that the empirical science of the formalist is often divorced from the values and intents of the experientialist. Model creation can come from a dialogue between all parties concerned, much like concurrent engineering, and thereby be acceptable to all parties concerned. Further knowledge is developed through the interplay of theory and practice, so that the process of model building and design is one of discovery as well.
As the value of broad experience, particularly in the early stages of model development, is recognized, and the limits of the computation paradigm in formalizing experience are reached, the importance of studies of meaning and the sign and symbol systems used to convey meaning are starting to be independently recognized.
Through this introduction, the topics of virtual environments, word processors, computer agents, design advisors, complex models, design as text, and shared memory have been discussed. Inherent in all of these is the issue of modeling: creating a synthetic environment through artificial representations. Such is the essence of design.
The theme of design-as-text can be more fully encompassed by the notion of design-as-literature, where literature encompasses an enormous library of speculations, theories, and techniques surrounding the written word. Some of these theories come direct from Aristotle while others have been fashioned in this century. With the advent of symbolic processing, more attention has been placed on the conveyance of meaning through the symbol - a subject with which literature studies have persistently been dealing.
Some of the issues seemingly arising anew within engineering design have been brought forth and discussed for more than fifty years within the field of semiotics. "Semiotics" is a study of signs. A semiotic sign consists of the signified meaning and the signifier representation. An example in the field of computers of a rediscovery of semiotics is the application of a similar sign system of the Class-Instance Pair (CLIP) model. The signified-signifier construct with an ordered pair consisting of an object and its icon is duplicated [Glinert & Gonczarowski 87]. The CLIP model, in this way maintains the association between the manifestation and underlying object.
The duplication of work within semiotics - the "re-invention of the wheel" - is inevitable as software engineers and computer scientists deal with the new abstractions of object technology, client-server architectures, distributing processing and virtual reality without reference to the similar systems that have been addressed within the virtual worlds of the arts and letters for centuries. [Glinert & Gonczarowski 87] even pause midway through the definitions of the CLIP model to reflect on the agreement users must make on the icon and its underlying meaning. This is the very contention of semiotics, that meaning is a social convention.
The concepts of semiotics and critical theory, previously relegated to debates within literary journals, now have platforms upon which to be implemented and explored. The computer can now act in both roles as workbench and implementation.