Computer models are critical tools for helping society understand complex phenomena. But more often than not, in the climate and hydrologic realms we oversell the ability of models to project the future and undersell their ability to help us understand critical physical processes including data issues. The act of model building should be a process akin to writing -- a deep intellectual exercise that reveals a variety of insights. Done properly, model building can be a valuable learning process that provides broad scientific and engineering insights including even what we can and cannot model. Done improperly, model building is a flawed effort solely focused on generating code and output where we fool ourselves about our ability to project the future. In the recent past scientists and engineers have succumbed to the understandable but ultimately flawed tactic of promoting modeling based on promised model outputs rather than on a more holistic process of using the effort to come to a deeper understanding of complex physical phenomena. Such modeling efforts will require more time and more introspection than is typical in many exercises, but the resulting understanding will be far more valuable for scientists, engineers and society.