Design is an activity that is subject to rational scrutiny but in which creativity is considered to play an important role as well. Since design is a form of action, a structured series of decisions to proceed in one way rather than another, the form of rationality that is relevant to it is practi- cal rationality, the rationality incorporating the criteria on how to act, given particular circumstances. This suggests a clear division of labor between the part to be played by rational scrutiny and the part to be played by creativity. Theories of rational action generally conceive their problem situation as one involving a choice among various course of action open to the agent. Rationality then concerns the question how to decide among given options, whereas creativity concerns the generation of these options.

This distinction is similar to the distinction between the context of justification and the context of discovery in science. The suggestion that is associated with this distinction, however, that rational scrutiny only applies in the context of justification, is difficult to uphold for techno- logical design. If the initial creative phase of option generation is con- ducted sloppily, the result of the design task can hardly be satisfactory. Unlike the case of science, where the practical consequences of enter- taining a particular theory are not taken into consideration, the context of discovery in technology is governed by severe constraints of time and money, and an analysis of the problem how best to proceed certainly seems in order. There has been little philosophical work done in this direction; an overview of the issues is given by Kroes, Franssen and Bucciarelli.

The ideas of Herbert Simon on bounded rationality are relevant here, since decisions on when to stop generating options and when to stop gathering information about these options and the consequences when they are adopted are crucial in decision making if informational over- load and calculative intractability are to be avoided. However, it has proved difficult to further develop Simon‘s ideas on bounded rationality. Another notion that is relevant here is means-ends reasoning.

In order to be of any help here, theories of means-ends reasoning should then concern not just the evaluation of given means with respect to their ability to achieve given ends, but also the generation or con- struction of means for given ends. Such theories, however, are not yet available; for a proposal on how to develop means-ends reasoning in the context of technical artifacts, see Hughes, Kroes and Zwart. In the prac- tice of technology, alternative proposals for the realization of particular functions are usually taken from ‗catalogs‘ of existing and proven reali- zations. These catalogs are extended by ongoing research in technology rather than under the urge of particular design tasks.

When engineering design is conceived as a process of decision mak- ing, governed by considerations of practical rationality, the next step is to specify these considerations. Almost all theories of practical rationali- ty conceive of it as a reasoning process where a match between beliefs and desires or goals is sought. The desires or goals are represented by their value or utility for the decision maker, and the decision maker‘s problem is to choose an action that realizes a situation that has maximal value or utility among all the situations that could be realized. If there is uncertainty concerning the situations that will be realized by a particular

action, then the problem is conceived as aiming for maximal expected value or utility. Now the instrumental perspective on technology implies that the value that is at issue in the design process viewed as a process of rational decision making is not the value of the artifacts that are cre- ated. Those values are the domain of the users of the technology so cre- ated. They are supposed to be represented in the functional requirements defining the design task. Instead the value to be maximized is the extent to which a particular design meets the functional requirements defining the design task. It is in this sense that engineers share an overall per- spective on engineering design as an exercise in optimization. But alt- hough optimization is a value-orientated notion, it is not itself perceived as a value driving engineering design.

The functional requirements that define most design problems do not prescribe explicitly what should be optimized; usually they set levels to be attained minimally. It is then up to the engineer to choose how far to go beyond meeting the requirements in this minimal sense. Efficiency, in energy consumption and use of materials first of all, is then often a prime value. Under the pressure of society, other values have come to be incorporated, in particular safetyand, more recently, sustainability. Sometimes it is claimed that what engineers aim to maximize is just one factor, namely market success. Market success, however, can only be assessed after the fact. The engineer‘s maximization effort will instead be directed at what are considered the predictors of market success. Meeting the functional requirements and being relatively efficient and safe are plausible candidates as such predictors, but additional methods, informed by market research, may introduce additional factors or may lead to a hierarchy among the factors.

Choosing the design option that maximally meets all the functional requirements (which may but need not originate with the prospective user) and all other considerations and criteria that are taken to be rele- vant, then becomes the practical decision-making problem to be solved in a particular engineering-design task. This creates several methodolog- ical problems. Most important of these is that the engineer is facing a multi-criteria decision problem. The various requirements come with their own operationalizations in terms of design parameters and meas- urement procedures for assessing their performance. This results in a number of rank orders or quantitative scales which represent the various options out of which a choice is to be made.

The task is to come up with a final score in which all these results are

‗adequately‘ represented, such that the option that scores best can be considered the optimal solution to the design problem. Engineers de- scribe this situation as one where trade-offs have to be made: in judging the merit of one option relative to other options, a relative bad perfor- mance on one criterion can be balanced by a relatively good perfor- mance on another criterion. An important problem is whether a rational method for doing this can be formulated. It has been argued by Franssen that this problem is structurally similar to the well-known problem of social choice, for which Kenneth Arrow proved his notorious impossi- bility theorem in 1950, implying that no general rational solution meth- od exists for this problem. This poses serious problems for the claim of engineers that their designs are optimal solutions, since Arrow‘s theo- rem implies that in a multi-criteria problem the notion of ‗optimal‘ can- not be rigorously defined.

This result seems to except a crucial aspect of engineering activity from philosophical scrutiny, and it could be used to defend the opinion that engineering is at least partly an art, not a science. Instead of surren- dering to the result, however, which has a significance that extends much beyond engineering and even beyond decision making in general, we should perhaps conclude instead that there is still a lot of work to be done on what might be termed, provisionally, ‗approximative‘ forms of reasoning. One form of reasoning to be included here is Herbert Si- mon‘s bounded rationality, plus the related notion of ‗satisficing‘. Since their introduction in the 1950s these two terms have found wide usage, but we are still lacking a general theory of bounded rationality. It may be in the nature of forms of approximative reasoning such as bounded rationality that a general theory cannot be had, but even a systematic treatment from which such an insight could emerge seems to be lacking.

Another problem for the decision-making view of engineering design is that in modern technology almost all design is done by teams. Such teams are composed of experts from many different disciplines. Each discipline has its own theories, its own models of interdependencies, its own assessment criteria, and so forth, and the professionals belonging to these disciplines must be considered as inhabitants of different object worlds, as Louis Bucciarell phrases it. The different team members are, therefore, likely to disagree on the relative rankings and evaluations of the various design options under discussion. Agreement on one option as the overall best one can here be even less arrived at by an algorithmic

method exemplifying engineering rationality. Instead, models of social interaction, such as bargaining and strategic thinking, are relevant here. An example of such an approach to an (abstract) design problem is pre- sented by Franssen and Bucciarelli.

To look in this way at technological design as a decision-making process is to view it normatively from the point of view of practical or instrumental rationality. At the same time it is descriptive in that it is a description of how engineering methodology generally presents the is- sue how to solve design problems. From that somewhat higher perspec- tive there is room for all kinds of normative questions that are not ad- dressed here, such as whether the functional requirements defining a design problem can be seen as an adequate representation of the values of the prospective users of an artifact or a technology, or by which methods values such as safety and sustainability can best be elicited and represented in the design process.