Reduced complexity modeling has had a long history in climate science. Model developers and students alike endeavor to use numerical models to better understand and simulate the climate system. Thus, these idealizations make for useful virtual laboratories. Recent efforts have continued to support the implementation of hierarchies in global climate modeling frameworks. These hierarchies can span from dry dynamical cores tests, to aquaplanet experiments, to comprehensive climate simulations. Idealizations allow for numerical experiments that target specific features of interest, while retaining the fundamental properties of the climate system, making them useful tools to test theory. Simplified approaches often require less computational resources than more realistic, CMIP-class, simulations. Despite these advantages, it is often challenging to support such hierarchies in operational modeling frameworks due to (1) the ever expanding coupling and complexity of models, (2) limited software engineering resources, and (3) the desire to represent the physical world with the finest resolution possible. As a result, modern model development, such as that for global cloud resolving models, often skips large portions of the hierarchy - prioritizing demonstrative simulations for short, superficial periods of time with limited model verification and minimal ensemble-size. Even at current comprehensive climate model resolutions, with the coupling of multiple Earth system components, the incentive to increase complexity with increasing computational resources limits the configurability of such models for scientific exploration.

The simplicity and flexibility of idealized models makes them powerful educational tools and enables imaginative science that pushes researchers to think in new and creative ways. Here, we demonstrate the continued need for operational models that re-emphasize model hierarchies for training and understanding with a review of recent advances. This includes (1) expanding idealized frameworks across Earth system components to include coupled processes and (2) more frequent use of these frameworks to help train the model development workforce in the coming decades.