Parallel Computation Models — Some Useful Questions

The Webster New Collegiate Dictionary defines the word model as, among other things, (1) a mathematical description of an entity; and (2) a pattern of something to be made. Scientists use the word in the first sense; their models are descriptive. Engineers use the word in the second sense; their models are prescriptive. Computer scientists, which are a little bit of both, use it in a hybrid sense: A computing model is an idealized description of real computers. This is the prevaling sense in areas with well established paradigms, such as sequential computing (with the Von Neumann computer paradigm). Computing models are also a prescription of how computers could, or should be built. This aspect is more important in parallel processing, where various paradigms are competing. Has complexity theory a role to play in the selection of such prescriptive model?