Deficiencies in sub-grid cloud and convection parameterizations distort basic modes of moist tropospheric variability in conventional global climate model (GCM) simulations on a range of timescales (e.g. diurnal, intraseasonal, ENSO). Experiments in the past decade using an alternative technique, the "Multi-scale Modeling Framework (MMF)" approach (i.e. embedding explicit models of convection in a GCM, a.k.a. "superparameterization") exhibit some intriguing improvements across this timescale range, as well as curious new distortions in the simulated mean state. This talk presents an overview of the MMF concept, its design philosophy, early history, and highlights from the past five years' research by the Center for Multiscale Modeling of Atmospheric Processes (CMMAP). CMMAP is a ten year multi-institutional US National Science Foundation Science and Technology Center that funds 100 researchers and 20 doctoral students (including the speaker). Its mandate is to explore and develop a portfolio of next generation climate simulation technology.