Introduction

A well-known problem for design rationale systems is to motivate designers to record the rationale for their decisions. One reason for this problem can be understood in terms of the ratio of cost for recording rationale to its perceived benefit. Systems must make recording rationale worth the effort required. Rationale systems that are shared by groups of people must also benefit the person who makes the effort to add rationale [Grudin 88].

Domain-oriented design environments integrate rationale systems with design tools, offering benefits to users by delivering rationale in support of the design process [Fischer,et. al. 91]. This delivery is accomplished by linking rationale according to the semantics of the design domain. For example, knowledge-based critic mechanisms watch over the designer's shoulder during the construction activity for violations of general kitchen design principles. When a violation is detected by the critic, a mouse-sensitive critique is displayed on the screen notifying the designer of the problem. Clicking with the mouse on the message causes the system to display argumentative information to the designer which explains the critique and possible ways to resolve the problem.

Good designers are characterized by their ability to transcend general purpose design principles. Their rationale for doing so is exactly the type of information that should be recorded and made available to other designers. In this spirit, mechanisms such as critics should not be thought of as enforcers of design principles. Instead, they should be thought of as informing designers about the intent of design principles and the conditions under which the principles may be broken.

Design environments reduce the overhead involved in adding to an existing collection of design rationale by locating argumentative information relevant to the designer's task at hand. The designer must input the rationale, but the system assumes responsibility for storing the rationale so that subsequent designers will access it when it is applicable. In this way, the construction tools of design environments provide an interface to underlying design rationale. As an interface to rationale, design environments determine what issues are easy to get information about and what issues are easy to add rationale about. The utility of these systems depends on their ability to support designers in accessing and adding information about issues that are relevant to the design task.

In this paper we outline the evolving artifact, or EVA, approach for designing systems that provide access to problem solving rationale within the context of the problem solving process. Our approach is intended to apply to all practices that require creative problem solving. It assumes that by carefully representing the problem solving practice of users we can create tools that facilitate the recording and sharing of important problem solving experiences.