The unified modeling language (UML) was developed to be a general-purpose design language. The initial version was a synthesis of three design methods. This approach is reminiscent of the development of the programming language PL/I, which was short-lived for various reasons, one of them being its overwhelming size. In these terms, UML resembles PL/I, yet there is a perception that it is fulfilling its role successfully.

To investigate the extent and nature of the use of UML, the author of this paper interviewed 50 software development professionals over a two-year period. The interviewee sample covered a wide range of developers within a variety of organizations of different sizes. The main question was whether UML was used and to what extent, including features, life cycle stages, and frequency. A scale ranging from use to non-use was identified based on the answers. The results show that 75 percent of interviewees do not use UML. A smaller percentage (22 percent) use it selectively.

Both this empirical study and previous studies surveyed by the author indicate that UML is used neither widely nor extensively. This situation brings to light a serious contradiction between the UML research community and its use by software developers. Through its activities, the UML research community claims implicitly that this design language has become universal, but most software developers do not appear to use it.

Even though this study is limited, it raises some major issues about the claims surrounding UML. Its evangelists need to address them and substantiate their claims. Language and software designers would benefit greatly by investigating why UML has not been used effectively. UML designers may also benefit from the experience of the programming language community, which realized long ago that the design of a general-purpose language involved too many conflicting tradeoffs and gave up on it. It may be time for the UML design community to reassess its approach.