A method for modeling tetrahedral networks structures is presented by which local rules are applied to the successive assembly of tetrahedral units which share elements (corners, edges, faces). In this way, even comparatively complex crystalline structures are easily assembled with surprisingly simple rule sets, but also topologically disordered structures may be generated by application of deviant rules in order to model the topologically-possible structures of network glasses. Optimization of aperiodic networks is accomplished by representation as spring-mass models. The advantage over hand modeling is that the adjacency matrix is immediately available from which we can extract topological parameters such as ring statistics. Furthermore, the atom coordinates are recorded during assembly so that obtaining information such as bond angle statistics is simple. The approach has been applied initially to SiO₂, SiC and Si₃N₄ tetrahedral networks in their several polymorphic crystalline manifestations and to several aperiodic silica possibilities.