Predicting the properties of new materials prior to manufacturing is a topic attracting great industrial and scientific interest. Property prediction for ceramic materials is facilitated by the periodic short- and long-range order of crystals. However, for glassy systems the lack of long-range order and the long time scales for relaxation greatly complicate the traditional modeling efforts. We show that the key for making progress has been to extract the key physics governing the macroscopic properties by using topological constraint theory, which was originally developed by J.C. Phillips and M.F. Thorpe around 1980. By further including the Gupta-Mauro temperature dependence of the constraints, the composition dependence of properties such as hardness and viscosity can be quantitatively predicted for oxide network glasses of industrial interest, such as borates and borosilicates. Moreover, the modeling approach enables a detailed understanding of the microscopic mechanisms governing macroscopic properties. Finally, we also present a phenomenological model offering an improved description of the composition and temperature dependence of the shear viscosity of multicomponent liquids, for which the existing analytical models currently do not apply.