The art and science of understanding complex, dynamical systems with memory

Why the Name?

#1. Cybernetics “master / slave” architecture

Norbert Wiener “invented” Cybernetics during WWII as part of the ongoing effort to develop specialized equipment capable of automatically tracking enemy aircraft and controlling air defense artillery. Anti-aircraft guns were at that time manually controlled by one or more people to assure a shell fired from the barrel would hit or explode close enough to the incoming target in order to destroy it or incapacitate the threat.

After Wiener would have sorted out the math, the engineers would replace the manual control wheels attached to the gearboxes of the gun’s two axes of rotation with a couple of electrical or hydraulic motors. The gun would then be ready to accept control signals from a ballistic computer tracking the target and assuring the fired shell meets with the target more often than when the gun was manually controlled.

This was the plan, but not how it actually happened for Wiener, because another, more simple, and more accurate predictor was selected for field trials, and Wiener was anyhow more interested in the wider applicability of the ideas he developed in his tracking and control math “system”, so “the scientific study of control and communication in the animal and the machine” was born from this failure to get a working solution for a real problem.

Because of this engineering origin, specifically the “intelligent” (human-like) control of other existing machinery, at the core of cybernetics is the distinction between control and controlled systems that is one of its main tenets. This separation is still very prominent in “modern” cybernetics and other (control) sciences that sprung from it.

A controlled system is usually some kind of simple, high-energy machine able to perform work (a gun, a steam locomotive, a car …). For the work to be useful, this powerful system must be controlled by a “smarter” low energy (“informational”) control system, able to monitor the operation of the controlled system in its environment and change (manage) its internal states to fulfill the goals of the system as a whole.

Kihbernetics does not subscribe to this “master/slave” architecture and considers systems as entities that are (as they are actually defined in cybernetics) “open to the exchange of matter and energy but closed to information from the environment”. Moreover, because such systems are “informationally tight” they are not only impervious to external control but are also unable to directly control other systems in their environment.

According to Kihbernetics, all a system can control is its own functions and states that may or may not result in an (externally) observable behavior within its environment. In other words, control, like growth, is only possible from the inside out. Any change in either the system itself or/and its environment is influenced, constrained but not controlled through a “structural coupling” between the system and its environment.

In kihbernetics, information, knowledge, and control are all intrinsic properties or capabilities of a system that are confined within the system’s boundaries and do not exist outside of it. The current state of the system depends primarily on the history of its previous state changes. Any “special” control parameter often used in cybernetics to control the state of the controlled system is in Kihbernetics treated just like any other “normal” input or disturbance from the environment that will be used at the discretion and in the way specified by the system.

And, yes, in Kihbernetics, the use of a thermostat as an example for a “control system” is strictly forbidden.😉

#2. 1st and 2nd order Cybernetics

Because of this “unnatural” dichotomy of control vs. controlled systems, M. Mead and H. von Foerster had to introduce “second-order cybernetics” in the ’70s as a reaction to the growing concept of autopoiesis defined at approximately the same time by Chilean biologists H. Maturana and F. Varela. According to this theory, it is impossible for an observer to fully know the external “reality” of the system.

In the good old cybernetic tradition of “divide and conquer”, the introduction of 2nd order cybernetics made a split between “observed” and “observing” systems, with traditional (1st order) cybernetics dealing with “observed” systems doing their trivial job of observing simple things and phenomena in their environment, while the “higher”, 2nd order, cybernetics was tasked with “observing the observer” and seeking to answer the question how is it that they affect that which is observed. Implicitly, observers are in cybernetics always thought of as cognitive living systems.

As a result, 1st order Cybernetics continued to work on “technical” systems and was gradually replaced by a new generation of exciting disciplines such as neural networks and artificial intelligence, while 2nd order Cybernetics got lost in the humanities and postmodernist discussions about how to deal with complex “social systems”. So another unnecessary dichotomy was born with the distinction between “hard” (1st order, engineering, simple, “command and control”) and “soft” (2nd order, “holistic”, complex, collaborative) systems sciences.

Kihbernetics does not make any such distinctions. Different kinds of systems (physical mechanisms, biological organisms, and social organizations) are all treated equally, as whole unities interacting with their environment, and the observer is just another dynamical system with memory “living” in and sharing the same environment with other dynamical systems. Some observers have an additional capacity of observing and reflecting upon their own cognitive processes. However, as much as the cognitive capacity of an observer may have a role in the discussion, in Kihbernetics, self-awareness is not a necessary condition for a system to be identified as an observer. Kihbernetics subscribes to the constructivist epistemology and recognizes that no observer, no matter how sophisticated, can totally grasp “objective reality”. At the same time, Kihbernetics does not refute the ontology of the fact that an independent common, and shared reality must exist as the phenomenal domain of interaction between observers and other (dynamical) systems and structures as the environment where to unfold their “praxis of living”.

In other words, in kihbernetics, if the tree falls and no observer is there to observe the phenomenon as it happens, two or more observers “seeing” the fallen tree can still agree on the fact that the tree was at some point standing there and agree on the most probable way it came to be laying there as it was observed.

#3. Cybernetics and “everything computer”

Another consequence of the period in which cybernetics was born after WWII, is its connection with everything computer-related. Although the declared mandate in cybernetics was always the study of systems in their generic terms of “communication and control” by identifying analogies and finding common rules governing their workings regardless of their substance (mechanical, living, or social), the then fascination with the emerging computer technology, “hijacked” cybernetics very early and this trend has never died, so that today we have a superabundance of “cyber” monikers, from “cyborgs” to “cyber warfare” and “cyber security”, all related to artificial computing technology. Even the traditionally conservative military succumbed to this “computerization” trend, so at some point, C3I became C4I by someone adding, you guessed, Computers (products) to the original (people and processes) functions of Command, Control, Communication, and Intelligence.

After a period of decline there seems to be in the new millennium a renewed interest in the use of cybernetics in other fields, so in a recent paper from S. Umpleby & all, “Advances in Cybernetics …” the authors identify three “Cybernetic models” with four associated “variables” to show the applicability of modern Cybernetics in the field of Social Sciences:

Whereas the social science disciplines create descriptions based on either ideas, groups, events or variables, cybernetics provides a multi-disciplinary theory of social change that uses all four types of descriptions.

Kihbernetics integrates all three cybernetic models (and more) in a standard, concise, and easy to understand Dynamical System Model (DSM) in which the “regulator” and the “regulated” are all part of the same self-organized system having the ability to observe and adapt to its environment due to the inherent reflexive learning capability that is a characteristic all dynamical systems with memory possess.

The DSM describes a generic state determined dynamical system with memory that can be used equally well to describe both complex “living” and “social systems” as well as conscious or not natural or artificial intelligence systems.

Kihbernetics is not an attempt to start from scratch by denying everything achieved so far in system sciences, including cybernetics and its predecessors. The main goal of kihbernetics is to return to the original intent of the science by reviewing and solidifying some of the basic tenets in cybernetics with the added benefit of the insight provided by old and new ideas in other disciplines and people working on similar issues both within and outside different systems sciences.

I chose the name “Kihbernetics” not only to do some justice to the pronunciation of the term’s Greek origins but also not to confuse it with the old cybernetics if (when 🙂 this novel idea starts getting some real traction in the systems thinking community.

Views: 26