SUMMARY

The broad aim of the proposed research is to contribute knowledge that enables improvements in how designers make decisions between system-level alternatives—i.e., between different potential system configurations and concepts. The ideal for designers is to have a sound and effective methodology for making such decisions. Although the general issue of decision making has received much attention in the design community, the reported methodologies are less effective for system-level decisions as formulated in this proposal. The specific aims of the proposed research are: (1) to construct a theoretical understanding of the problem based on accepted decision theory and (2) to demonstrate a solution methodology informed by this understanding. A basic premise of this research is that decision theory provides an adequate basis for system-level decision making, but that approaches for modeling system-level alternatives are deficient. The principal research question is: how should designers model system-level decision alternatives in order to support sound and effective decision making? The general hypothesis is that designers can model a system-level alternative compositionally using predictive tradeoff models of its subsystems. These models abstract the capabilities of a subsystem in a way that is independent of implementation details. This differs from the classical approach of using analysis models to map from design parameter space to a tradeoff space, which is undesirable for system-level decisions because designers typically lack adequate parametric descriptions of their alternatives. The principal measure for success is whether or not the modeling approach allows designers to reach decisions in a manner consistent with accepted decision theory. This is evaluated using a combination of logical argument and empirical comparisons with trusted decision methods. A secondary consideration is whether the benefits of the approach are likely to outweigh the associated costs.