acm - an acm publication


Whatever happened to cybernetics?

Ubiquity, Volume 2008 Issue February | BY Ross Gagliano , John Gehl 


Full citation in the ACM Digital Library

Has the discipline of cybernetics been unable to recognize and respond to appropriate "midcourse corrections" and in the process had its destiny imposed by external "turning points"? (A mid-course correction is an endogenously determined action as in the classical "sense-processact" sequence, whereas a turning point is simply a reaction to that which is exogenously imposed; i.e., "being overcome by events."

Has the discipline of cybernetics been unable to recognize and respond to appropriate "mid-course corrections" - and in the process had its destiny imposed by external "turning points"? (A mid-course correction is an endogenously determined action as in the classical "sense-process-act" sequence, whereas a turning point is simply a reaction to that which is exogenously imposed; i.e., "being overcome by events."

Over the past 60 years there have been a number of events that have had an impact of one form or the other on cybernetics. Probably the very first was based on Wiener's definition of the discipline in the late 1940's as "the science of control and communication in the animal and the machine." But since then there have been more than a few attempts to elaborate on this seemingly simple, yet obviously complex and multi-disciplinary, topic. Further, there have many investigations into other equally intriguing and related concepts such as system, feedback, regulation, and intelligence.

"Feedback" and "regulation" have been successfully explicated for thermostats and autopilots, or "black boxes" whose relationship to traditional cybernetics (or control theory) is fairly straightforward. Clearly, it is a relatively simple matter to design and build a box to guide vehicles from point A to point B in space - whereas it has been far more difficult, and not yet entirely solved, to produce one that will guide a vehicle between two locations on the ground. One of the more perplexing aspects of this distinction between the two types of devices is the notion of "system."


From the very beginning, there has been a misconception that cybernetics implies that control in the animal and the machine are the same. As a fundamental premise, it is questionable, and there are good arguments that show that it cannot be true. A second reason why the concept of control across animal and machine has become blurred is the persistence, especially by general systems theorists about a single notion of "system." Suffice it to say that systems can be classified as of two varieties: natural (e.g., the solar system) versus manmade (e.g, a satellite communication system).

Disagreements over definitions, however, really beg a larger question of whether there should be only one science. As in the case for "systems," the real issue becomes whether there are substantial differences between the "sciences;" that is, natural sciences versus the "design" sciences. Papers presented at meetings of the Society for General Systems Research, for example, continually made such a distinction as a means to characterize turning points in public sector performance. Thus, one might surmise, that there has been a continual failure in the grand scheme of things of importing artifacts and tools from the natural sciences into the design sciences. A similar case could likewise be made for cybernetics because it spans the spectrum of sciences. Further, cybernetics is very closely tied to modeling, or the process of creating representations - a human activity necessary to understand, imitate, communicate, reproduce, and control. Moreover, modeling is not just an intellectual activity, but also a fundamental criterion by which progress in the natural sciences is tracked.


Political, cultural, and even professional issues also have tended to obfuscate cybernetic concepts, often with scientific and general systems overtones. After Wiener's cybernetics book was published, its ideas spread quickly around the world. They were, obviously, well received in the Soviet Union where Wiener made several trips, the most important of which was in 1960. After that, cybernetics began to play, at least from the perspective of their planned societies and economies, an extremely important role in political, economic, and cultural evolutions and counter-evolutions - i.e., East vs. West.

In retrospect, there have been some very interesting albeit diametrically opposite effects. In the East, cybernetics became a salient feature in communist pedagogy, as institutes were established with cybernetics as both the core academic discipline and the institutional political dogma. Meanwhile in the West, there has been a corresponding decline in cybernetics witnessed by fewer university course offerings and an ever-decreasing number of publications and forums.


Three significant research events have been suggested as "turning points": the perceptron debacle, the uncertain legacy of systems dynamics, and the creation of the fuzzy set phenomenology.

The Perceptron Debacle. Research into what was originally called the perceptron began in the late fifties based on work by Rosenblatt. He attempted to demonstrate how a collection of neural-like nodes could be connected and fired electrically to simulate the process of "learning patterns"- in a manner that real neurons were thought to behave in the living brain. After years of successes and failures, this work was officially denounced in a government review that ostensibly evaluated the relative merits of artificial neural networks in terms of the then new rule-based expert systems approach. Needless to say, that led to the demise of the perceptron because of its "computational difficulties," and, it appears, discrete element modeling. This gave rise to the prominence of expert systems within AI, an area that for the next 20 years received the preponderance of support.

Systems Dynamics Debate. The second missed opportunity has involved a continuing controversy over the proper use of certain closed-form mathematical expressions. Specifically, systems dynamics employs coupled sets of linear difference equations to model urban, regional, and global dynamics. Interestingly, there was an actual "debate" between two highly respected academicians, but not conducted face-to-face. They were the game theorist Martin Shubik of Yale University and Jay Forrester of MIT who was one of the Whirlwind developers prior to becoming the "father" of systems dynamics. This "debate" started with Shubik's review of Forrester's book, and continued in a series of letters to the editors of the journal Science. A basic contention in their dispute was whether the world, or some portion thereof, could be modeled as an electrical circuit, or a so-called "LRC" (for inductor, resistor, capacitor) system.

Fuzzy Sets, Logic, and Applications. Lofti Zadeh proposed an alternative mathematical set construct that he called "fuzzy" that subsequently encountered enormous resistance, particularly as it was extended to fuzzy logic and operations. Nonetheless, the culmination was the fabrication of a fuzzy chip. Although there is still be a great deal of controversy surrounding the theoretical relevance of some of these ideas (moreso in the U. S. apparently than in France, Bulgaria, or Japan), many practical applications appeared. Besides providing an ingenious solution to the classical 'cart-pole' problem, also called the 'reverse pendulum' or 'broomstick balancer' problem, commercial products employing fuzzy chips are now marketed including self-leaming, self-organizing, and self-controlling room vacuum cleaners, dishwashers, and camcorders, as well as other industrial process control devices.


Three areas of research that could assist cybernetics are the discrete element modeling approach; " system/organization"-type dichotomies; and the notion of a " market" as a valid modeling construct.

Discrete Element Methodology. Concurrent with the re-emergence of ANNs has come a realization of a coherent discrete element methodology (DEM). Such a methodology naturally integrates data structures and algorithms for representing complex processes while facilitating their extension into many diverse applications. Key ingredients of the DEM include: a collection of individuals, their associated attributes, strategies, and responses. Subsequent responses are based on, and likewise determine, which individuals monitor which other individuals (who "sees" whom), the disposition of individuals to modify their response or strategy ("sensitivities"), and the transition of individual responses (rules of strategy). The sequence of aggregate values (patterns of distributions) allows a form of description of the model as state transitions.

Because of its discrete elements, the DEM has inherent strengths as a consistent manner to correlate aggregation and disaggregation; a natural integration of mathematical components and modeling tools (both deterministic and stochastic); and a facility to trade-off detail and sophistication between mathematical modeling and computer simulation. Several other examples have been developed and tested that indicate a range of applications from neurocomputing to physiology to gaming. Disparate concepts such as "learning" in the individual (micro) and "mutations" in the aggregate (macro) can then be explicated through resultant "shifts" in patterns of model relative frequency histograms.

System/Organization Dichotomy. To effectively utilize the DEM, the notion of system can be dichotomized at least in the following way. Distinctions can be between an "organization" and a "system," and that was previously illustrated through an "orchestral dual." A 'system' was related to a classical orchestra with a conductor, a musical score, and the standard notion of errors. This was distinguished from an 'organization' that was related to a "jazz orchestra" that has no conductor, no written score, and a mutative notion of errors.

The Market Concept. Besides an appealing approach by the DEM, the study of the brain indicates that it can be modeled as an economic entity because it operates on the "principle of least effort," or cost. There is some inspiration that others have felt strongly about the notion of a market such as the late Friedrich August von Hayek who won the Nobel Prize in economics. He proposed a form of order based on a spontaneous market system that he called "catallaxy," implying a combination of "exchange" and "turning an enemy into a friend." Concepts such as cooperation and competition are critical in modeling human "systems."


Using the system/organization dichotomy, control can be characterized as either 'centralized' or 'decentralized.' Centralized control implies that the control functions are implemented by a single entity (regulator, governor, or controller). When several entities cooperatively carry out these functions, then the control is decentralized. Decentralized control has been modeled in election paradigms (i.e., voting algorithms), as well as through a competition in markets. In some control paradigms, the terms 'host' (or hosted) is used for the centralized case and 'hostless' for the decentralized case. In the hosted case, resources are allocated based on priorities or schedules whereas in the hostless case they are allocated through "self-organizing" algorithms.

We have looked at the problem of resource allocation in several environments, especially decentralized control for computing environments. We have also investigated various types of markets: sealed-bid auction, barter exchange, and challenge protocols. Through simulation, we have provided "intelligence" to computing tasks such that they "bid" through auctions, "exchange" by bartering, or "share" through challenges to gain access to needed resources. Admittedly simplified, such computing configurations can be generalized to larger sets of hardware, operating systems, and network architectures.


We question whether, during its evolution, cybernetics has itself failed to be self-correcting. As a result, some commonly held beliefs have been challenged about science in general, and the cybernetic discipline in particular. What has been indicated is the possibility for a "new cybernetics" or a "science of science." A major example described in the last reference is a market model based on a hostless control of resources.

REFERENCES Ashby, W. R. Introduction to Cybernetics. London: Chapman and Hall, 1956.
Bertalanffy, L. von. Problems of Life. New York: Harper, 1960.
Forrester, J. World Dynamics. Cambridge: MIT Press, 1971.
Gagliano, R. A. "An Approach to Solving the Systems' Dilemma," Proceedings of the 11th Annual Pittsburgh Conference on Modeling and Simulation, 1273-1276, May 1980.
Gagliano, R. A. and K. S. Lebkowski, "Cybernetics and a General Theory of Organization," presented at the 4th European Meeting on Cybernetics and Systems Research, Linz, Austria, March 1978 and published in Progress in Cybernetics and Systems Research, Vol. VII, Hemisphere Publishing Co., 143-148, 1979.
Gagliano, R. A. and A. P. Schwartz, "Systems, Models and the Jazz Orchestra," presented at the International Congress on Applied Systems Research and Cybernetics, Acapulco, Mexico, December 1980 and published in G.E. Lasker (ed.) Applied Systems and Cybernetics, Pergamon Press, 687-691, 1981.
Gagliano, R.A., Fraser, M.D. and M.E. Schaefer, "Auction Allocation of Computing Resources," Communications of the ACM, Vol. 38, No. 6, 88-99, June 1995.

Ross Gagliano is a retired professor and co-founder of the Computer Science Department at Georgia State University. Previously, he was a senior researcher at the Georgia Tech Research Institute.

John Gehl, the editor-in-chief of Ubiquity, was for many years on the staff of Georgia Tech, and for four of those years headed the Office of Computing Services. Starting in the 1990s he co-created, with Suzanne Douglas, such online publications as NewsScan Daily and Innovation Weekly.

Source: Ubiquity Volume 9, Issue 7 (February 19, 2008 - February 25, 2008)


Leave this field empty