In this chapter we have defined ISD, and described methods and tools. First, for the purposes of metamodeling, methods were seen to consist of different types of method knowledge. This analysis focused on method knowledge related to modeling techniques, i.e. on the conceptual structure and notations. Thus, we excluded other aspects of methods and their development. Second, we have described the relationship between modeling tools and methods: the method-tool companionship. This allowed us to show what type of computer support is needed to develop tool support (i.e. abstraction, checking, form conversion and review).

Third, we discussed method use through the notion of method paradoxes. The analysis of method use revealed that the applicability of existing methods is not at all clear, because many ISD organizations do not use the available standard-like methods at all, and have developed their own partially or completely new methods. As a result, the IS research community must admit that we do not know well enough how methods are actually used in development situations, and how important the role of methods is in the success (or failure) of ISD efforts. These paradoxes led us to refine the currently dominating view of methods: we defined methods to be situation-bound instead of universal and standard. We acknowledged that a method is not the sum total of ISD knowledge, as much knowledge about ISD is tacit and can not be provided as readily applicable routines. We emphasized expertise and learning, and viewed methods as evolutionary.

Based on the IS research literature, there appear to be at least three possible ways to research method use. The first is to continue the widely followed research approach to develop new situation-independent and universal methods, compare them conceptually (e.g. frameworks), and use them in cases. However, this approach, despite its use in multiple studies, has proven to be inadequate for resolving problems related to the wider acceptance of methods. The second option is to pursue comprehensive empirical studies on methods in realistic environments (e.g. as proposed by Wynekoop and Russo 1993). Although this proposition is basically correct, it is not a realistic approach for today’s organizations. First, they can not stop their ISD efforts and wait for the results. Second, the results of these empirical studies can become obsolete even before they are ready, because of the rapid evolution of the business world and technology. For example, there is not much empirical evidence on the usefulness of object-oriented methods, although this is one of the challenges for ISD in many organizations today. Similarly, there is a paucity of research examining the usefulness of metaCASE tools (Tolvanen et al. 1996).

The third option is method engineering: to focus on mechanisms that support local method development and use. Although many companies are “rolling their own”, using local, in-house methods, method development seems to be carried out in an ad-hoc manner by selecting tools and methods on a trial-and-error base. Organizations do not have any principles to guide ME efforts: selecting and constructing methods for particular needs, checking the completeness of methods, or organizing method development efforts. Moreover, organizations face problems in finding and developing tool support and collecting experience of method use. All these reasons motivate the development of systematic principles for ME. In the following chapter, we shall describe approaches or strategies for method selection, construction, and tool adaptation.