Thomas A complex set of observations and questions. We offer some responses as follows: Variables can be usefully thought of as the distinctions which may assume different values of the domaim over which they are defined. Think of the relationship between algebra and arithmetic. The variables of algebra represent different domaims (typically in the number system) and may assume the value of any part of their domains. This is analogy but useful analogy. We will not pursue here the many distinctions with this general area but do recommend that you do (bound variables versus unbounded variables...). State, BTW, is a rigorously defined notion in many fields - the origin in NLP of the use of this variable is Automata theory where it is the instantaneous description of the machine (read organism for NLP) of the system under study. In NLP, state is (in our opinion and work) must usefully defined with reference to physiology - see the section on the Chain of Excellence in the New Code portion of Whispering. Your comment that state in cybernetics is defined as the output that results given any particular input leaves out the critical issue of the state of the system upon receipt of the input involved. In Automata theory, for example, the rules that defines what the machine (organism) will do (output) is dependent not only on the input but crucially on the state the machine is in upon receipt of that input. The commom meaning of state is not too terribly off the mark as well - as you note. You offer several sets of examples of possible variables - the meta model, the representational systems - indeed, your guesses are accurate within the sets so defined. What we were referring to in Whispering is the set of variables commom across all those models of NLP. These are obviously higher level variables. The verb "to model" and the substative "model" have a huge coverage in common usage, even in the somewhat more restricted usage in scientific endeavors. In Whispering, we worked very hard to offer an explicit representation of what we perceive distinguishes NLP modeling from these other variations. Thus your example, of using the term "model" to refer to your figuring out in what specific sequence your subject model uses representational systems to successfully do some computation coincides with common usage and patentedly NOT what we propose for NLP modeling. Do not be alarmed by the immense gap between the variables alluded to in our work and their deploymen in the New Code games. You have not been provided with adequate examples and most certainly adequate experience to discover that connection. We did not (as we stated in the text) attempt in Whispering to elucidate either the set of design variables we referred to nor the structure of New Code games. All the best, John and Carmen