Design is often cast as a process of generating and testing. The evaluation of potential design options effectively steers the subsequent generation of new designs. In turn, the options generated help to shape the evaluations used to choose from among them. This coupling between generation and evaluation is a vital part of the design process and is perhaps the most significant way in which decision-based design can be differentiated from decision-making: Designers are charged not only with selecting the best design options, but also with generating the options from which they must choose. Human designers use abstraction throughout the design process, guiding the design through a series of representations: the social constructs of the customer∕environment from which the design need springs, textual∕graphical descriptions of this environment, “black box” functional descriptions of how a design will interact with the environment, detailed descriptions of interacting components that accomplish this interaction, spatial layout of these components, detailed geometry of connections among them, manufacturing∕assembly plans for each part, etc. Designers develop and evaluate design alternatives at each of these levels of abstraction before moving to the next; DBD can help to formalize the decision-making within and across abstraction levels, at some points removing candidates from the set of options, at others selecting the best one. In this chapter we lay out a methodology for generating design alternatives across these design abstraction levels and build on decision-making concepts from prior chapters to help control this process: probabilistic design modeling, value functions, expected value, decision-making under uncertainty and information value theory. Each of these is reframed for the task of conceptual design. In addition, a new concept-design freedom-is introduced to help designers evaluate the impact of their decisions on the diversity of the design candidate set. DBD over multiple abstraction levels is not a simple process of generate and select; it is a set of cascading decisions that refine both the abstract design and the initial requirements used to evaluate it.

People routinely compare similar instantiations of objects or classes of objects in which they are interested, whether they are digital cameras, colleges or potential spouses. Designers are routinely charged with, first, generating or creating a set of alternatives, and, second, choosing one of them as the most preferred. Thus, designers routinely rank objectives, design attributes and designs using decision-centric methods and techniques, many of which are variants of the standard pair-by-pair pairwise comparison [1–8]. The common underlying concept in these methods, now all embodied within the rubric of decision-based design (DBD), is that some part of design thinking can be represented and modeled as a decision-making process that is aimed at addressing the need for a rational way to choose among alternatives. But, which part of the design process—or perhaps how much of the design process—is decision-making? We attempt to address such questions based very much on a prior analysis [9], some of which was also detailed in [10], and augmented in part by a recent inquiry into design thinking, teaching and learning [11].