This paper examines GTE from an authoring perspective. It describes how GTE may be used to create a piece of courseware and reviews the potential difficulties that authors face when using such a tool. It summarises two studies that were carried out to investigate how GTE is used by actual authors, and concludes that GTE designs are hard for authors to test and evaluate because of the complexity of such designs. It finishes by arguing that authoring tool developers should give greater consideration to supporting the task of authoring, and that such consideration should occur at all stages of a new tool''s development.