Requisite Variety

No system, if it excludes an environment, can muster the, as W. Ross Ashby put it, "requisite variety" that would be necessary to establish a kind of matching, a kind of point-by-point relationship between System and Environment.

The system does not have the capacity to put its own state on top of everything that happens in the environment, to hold its own operation against it, either to promote or to prevent what happens, but it has to bundle or even ignore, it has to muster indifference or create special facilities for complexity management. In the discussion this became the formula that the system must reduce Complexity, and this on the one hand in relation to the environment and on the other hand in relation to itself, if it wants to create planning instances or rationality agencies within itself.

In the history of science, this problem can be traced back to the functionalist psychology of the 1930s. At that time, the limits of the input-output model or the stimulus-response model, as it was then called, were recognized. It was seen that one could not simply work with a one-to-one order of stimulus and response, but that between them there was something that provided the transformation and could not be reduced to a mathematical function.

LUHMANN, Niklas, 2004. Einführung in die Systemtheorie. 2. Aufl. Heidelberg: Carl-Auer-Verl. Sozialwissenschaften. ISBN 3-89670-459-1, p. 168–169

and p. 178: Another point that has always fascinated me concerns the question of what happens when two complex systems get involved with each other, when they are coupled or interact and do not have the ability to duplicate the complexity of the other in their own system, that is, do not have the "requisite variety" that would be required to map another system into themselves.

According to a thesis of Donald McKay, a Scottish cyberneticist or information theorist, Freedom arises under these conditions: Even if complex systems were machines and completely determined, each system would have to assume that the other can be influenced, i.e. react to signals, and this not in a way whose determinacy could be calculated in the system itself, but precisely in a way that is unpredictable.

Therefore, you have to sweeten the information, so to speak, you have to offer incentives that you believe or know from experience that the other systems will get involved, that they will cooperate voluntarily, based on their own preferences, or, if you want to rule that out, that they will not cooperate, that they can therefore decide and are not already through-determined systems that do what they do anyway.

I don't know what you think about that. In any case, it's a thing to think about, and it resolves a bit of the old discussion of whether the world is either determined or indeterminate. We are then again dealing with a paradox of the distinction: The world is indeterminate because it is determined, but this not centrally, but locally.

"Requisite variety" (Ashby), i.e. the Possibility to assume as many States as what it [the planning entity] plans or as the world outside. It must reduce Complexity. (p. 236–237)

W. Ross Ashby, Requisite Variety and Its Implications for the Control of Complex Systems. In: Cybernetica 1 (1958), S. 83-99.a