The Role of Expertise in the Evaluation of Computing Architectures supported by the literature in other (non-MIS) business areas. Rather than assume expert-novice differences in performance and seek to establish a cause by studying the process, future work in MIS should first establish performance differ- ences before delving further. This is especially true if the task under study is unstructured and subjective. Second, our findings contribute to the business literature in general and add weight to the growing body of evidence that while process differences do exist, performance differences between expert and novice managers for several tasks are not detected. This suggests that several tasks in the business domain are not structured and may not have a core set of domain-specific principles that are better understood by experts than novices. Third, we confirm findings from the psychology, business, and MIS literatures that experts adopt a top-down, heuristic methodol- ogy when performing a task and draw on earlier message to vendors to focus on developing higher quality software and emphasize that in their com- munication to MIS managers. Future WorK It is clear from our work that, as in other business areas, the process-performance paradox is alive and well in the MIS domain. In future work, it will be very interesting to understand which managerial tasks actually lend themselves to expertisethat is, what the tasks are where the performance of expert and novice managers differs. While it is clear that performance differences will exist for highly structured tasks such as programming and systems development, as the IS manager moves higher up the ladder and the tasks get progres- sively more unstructured, which tasks lend them- selves to expertise? Of equal importance will be