Ok ... so if its to be a generalised model, I suggest we start a new thread.
As a starter, one way of looking at this is from the basics of Complexity Theory ... start out considering the different dynamics we're likely to encounter in our chemical complexity model: static, dynamic, evolving, self-organising. All parameters would have to be variables, which change over time, at different rates.
Global measures would need to apply in all fields (as an assumption), along with others covering unpredictability, non-equilibria, causal loops, and openness.
Some objectives wouldn't go astray either. Eg: explanations of the structures, relative complexity, control methods, constraints, chemistry and physics laws, and they have to be computable (quantified).
Throw all that into an iterative algorithm all at once, and see what pops out, eh ?