Recent reviews of studies on the global information and communication technology (ICT) sector's current and future direct climate impact have provided general overviews and identified contradictory results regarding future impacts. So far, however, calculation approaches and strategies for future scenario construction behind these results have not been scrutinized in more detail and reasons behind contradictory results remain unclear. The aims of this study are (i) to identify reasons behind contradictory future projection results in previous studies of the global ICT sector's direct climate impact, (ii) compare these results with a recently published estimate for 2020 and (iii) to identify critical aspects and related practical challenges in these assessments. This was done via detailed reconstruction of future projection results combined with quantitative analysis of these results, focusing on model and parameter uncertainties. The results show that several of the reviewed studies apply strategies for future scenario construction that in practice do not allow decreasing emissions. Other reasons behind contradictory results are mainly due to model uncertainties, e.g. due to different scopes, but parameter uncertainties, related to e.g. number of end-user devices, are also significant. The comparison of future projections for 2020 with an estimate for the same year shows that projections tend to be significantly higher, e.g. 120-620% higher for global data center use phase emissions. The results from this study highlight the need for practical guidelines on how to conduct such studies, especially for identified critical aspects, such as limited inclusion of fast-growing and emerging digital technologies in previous studies.