The development of quantitative models of landscape evolution requires a rigorous assessment of how well they represent natural system dynamics. For some locations and processes, the models have been shown to provide reasonable representations of landscape dynamics over a range of timescales; however, in a number of cases, rather straightforward tests suggest our simplified approaches may require further development. I discuss some of these lingering issues in quantitative geomorphology with a focus on the ubiquitous ‘stream power model’ and show how different approaches offer potential solutions to implementation in landscape evolution models. I will show examples focusing on how climate and lithology are represented in landscape evolution models as well as discuss the limitations of representing a river channel with a 1-dimensional model. I summarize by suggesting that an iterative approach of numerical predictions, rigorous field assessment, and adjustment of theory is needed to fully capture and understand the dynamics of landscapes.