International Encyclopedia of Systems and Cybernetics

"A set of computer aided methodological tools for dealing with the relationship between wholes and parts, i.e. between overall systems and their various subsystems" (G. KLIR, 1991, p.141).

KLIR, who created and developed this methodology, states: "At the highest level of generality, reconstructability analysis may be said to be concerned with two complementary problems:

"1. The reconstruction problem: To what extent is a given system description inductively inferable from descriptions of subsystems deductively inferable from it? (1958, p.6).

(George KLIR, states this point in the following terms: "The aim of the reconstruction problem… is to determine the smallest possible subsystems by which a given overall system can be adequately represented) (1990, p.33).

2. The identification problem: To what extend do the descriptions of a given collection of systems determine (deductively) the description of a single system of which they are each subsystems".

"While recognizing what is now commonplace in science, i.e. that "the whole is greater that the sum of its parts" i.e. that it is (usually) not possible to infer from detailed knowledge of subsystems similarly detailed knowledge of a system of which they are parts, practitioners of R.A. seek decompositions of systems into subsystems such that as great a simplification as possible is achieved while keeping the loss of information regarding the original system within acceptable limits". Hence the characterization by CONANT (1988) of R.A. as "a form of enlightened reductionism" (1958, p.6).

Says KLIR: "In general the term "system" is used in R.A. for a set of variables together with some characterization of the constraints among these variables. Each variable of a system is viewed as an abstraction of some real-world attribute. It is associated with a finite set of states (values) each of which represents a class of appearances of the corresponding attribute" (1990, p.34).

These values or "states" must be measured.

KLIR adds: "In all these measures, the amount of information is measured in terms of the amount of reduced uncertainty. The more our total ignorance (full uncertainty) regarding states of relevant variables is reduced by the given system, the more information the system contains."

"… A discovery that a system can be represented by a specific set of subsystems may provide the investigator with some knowledge that is not avalailable, at least explicitely, in the corresponding overall system. For example, the subsystem configurations may give him information about causal relationships, the significance of the individual variables, the strength of dependencies among them, etc…" (p.35).

One could add that the whole is not only greater than the sum of its parts, but is something else, in some sense, more than the parts and in another sense less than their sum.

The totality masks many specific characters of the parts, some of whom may -or may not – be significant for the whole. One may surmise that the significant ones are those which intervenes in the interrelations among the parts.

However, up to some point, these interrelations remain occult when we look only at the global integrated system.

Thus, discovering them seems to be the mainpoint of R.A. Moreover they can be discovered only through their action, which can be measured through the varying states of the subsystems or parts, interconnected in specific and exclusive ways.