International Encyclopedia of Systems and Cybernetics

M. BUNGE proposed an epistemological evaluation, not of "a" General System Theory, but of the different General Systems Theories in existence, as for example "… the statistical theory of information, game theory, control theory, automata theory, and even the general Lagrangian framework, the general classical theory of fields, and the general quantum theory of fields (misnamed "axiomatic field theory") (1977, p.32).

He considers that: "All of these theories are phenomenological, i.e. are mechanism-free… Hence, they can describe the behavior of certain systems but not explain how they work. Moreover they are stuff-free – i.e. they make no detailed assumptions concerning the nature of the components of the systems concerned. Whence their extreme generality" (Ibid).

As a result, G.S.T. have no specific predictive power, unless complemented with "extra knowledge concerning the composition" of the specific system under consideration.

BUNGE gives an example, inspired by the case of the TURING Machines, "which being equipped with infinitely long tapes, are strictly speaking unrealizable".

As a result, "… while automata theory is applicable (by specification) and moreover guides much of advanced engineering design and even some psychological research (SUPPE), it is irrefutable. It is not even confirmable in the traditional manner of predicting and checking: the theory makes no specific predictions, it prohibits hardly any event, and it suggests no experiments other than Gedankenexperimente" (i.e. Thought experiments as proposed by EINSTEIN) (Ibid).

In short, concludes BUNGE, G.S.T.s are" empirically untestable".

He admits however that they can be bettered: "In fact, G.S.T.s are corregible if not exactly refutable in the light of empirical evidence. To begin with, they can be improved upon formally. i.e. logically or mathematically… Or they can be made more complex in an attempt to fit better their intend referents" (Ibid., p.33).

While not confirmable in traditional manner, they can be, according to BUNGE, confirmed conceptually and also practically "Thus general network theory is confirmed conceptually by being shown to capture the traits common to all nets, whether physical or informational. And it is confirmed practically by being used in the design of nets of some kind" (Ibid., p.34).

Without being the following truly a confirmation, it should also be noted that network models make sense and can be used for acquiring a better understanding of some systems in which very numerous elements are connected in a semi-permanent and more or less probabilistic way, as for example social systems.

BUNGE states moreover that "G.S.T.s are so many frameworks that must be adjoined specific models of the system of interest before they can be of any use. For example, in the case of information theory we must be able to identify at least the sources of information and of noise, the channel, and the receiver; and we must have a code for the system of signals" (Ibid).

(Let us observe, by the way, that BUNGE, no doubt, speaks here of SHANNON's communication theory)

"In other words, a member of the G.S.T. class becomes a specific theory of the standard type, perhaps a scientific theory, when enriched with specific information concerning the system to which it is applied".

And "G.S.T.s are confirmed a) by fitting all families of specific theories (conceptual confirmation) and b) by helping in the building of specific theories that are tested the classical way (indirect confirmation)" (Ibid).

Finally: "G.S.T.s are generic frameworks helping one to think of entire genera of entities in a variety of domains, from biology and psychology through hardware and human engineering to city planning and politics. True G.S.T.s solve no particular problems without further ado – but on the other hand they help in the discovery and formulation of new problems and they clarifiy basic ideas in all fields of inquiry" (A. RAPOPORT, 1972, p.74 – as quoted by BUNGE).