Two concepts of system

Главная / Публикации / Two concepts of system

Two concepts of system

G.P. Shchedrovitsky Two concepts of system // J. Willby, ed. Proceedings of the 46th Annual Conference of the International Society for the Systems Sciences, Asilomar, CA, July, 2002 (Перевод работы Г.П. Щедровицкий Два понятия системы // Труды XIII Международного конгресса по истории науки и техники. Т.1а. М., 1974)


Seminar on General Methodology


When one characterizes a “system”, one asserts that it is a complex unity in which one can distinguish constituent parts, or elements, as well as a network of connections, or relationships among the elements. It is as if through this definition, we envision the object comprised of elements and connections among them. What we see is an ontological picture adopted by the system approach. This ontological picture stands for all the procedures and methods that we apply to various symbolic constituents of scientific disciplines, representing their subject-matters as systems. In other words, the ontological picture of a system represents processes of system analysis, but it does not represent the functioning of a system itself or its parts. It is not surprising that traditionally researchers have reduced system processes to structural and parametric characteristics, thus accepting that a system is determined by its structure rather that by its processes. The situation has reversed with the advent of system design. System designers start with a specification of system product or goal and corresponding processes or activities. Only then do they decide on materials that should realize the processes. Thus system processes effectively determine system and its boundaries, including its material structure. This new approach gradually infiltrates contemporary science proper. Modern system analysis can no longer be confined to traditional analytical procedures and ontological representations. It requires a new analytical method and new types of representation. According to the new approach, representing a complex object as a system means representing it in the sequence of four categorical layers: (1) a process of a certain kind, separated from any material realization; (2) functional structure; (3) material arrangement; and (4) morphology. The main task of the new system analysis becomes establishing a correspondence among these layers for a particular complex object. One advantage of this new concept of system is that it allows determining correspondences between system processes of any kind (including evolution) and material arrangements and structures.

* * *

When we speak of a system today, whether referring to conceptual content or to an object, we usually mean a complex unity in which one can identify constituent parts, i.e., elements, and a network of connections or relationships among them. Sometimes, one adds dependencies among connections as well. Behind this definition, we virtually see the object as consisting of elements with connections among them. What we see is precisely the ontological picture adopted by system approach. This ontological picture stands for all the procedures and modes of operation that we employ at various levels of investigation when reproducing objects as systems (Schedrovitsky, 1964; 1965). These procedures and modes must be revealed if we wish to determine the objective content and the logical structure of the concept of system.

There are at least three groups of procedures underlying the ontological picture of a system as defined above. The first group includes two procedures: decomposition of an object into its parts, and composition of a whole out of the parts.  Composition is usually performed with the help of connections added for this purpose.  Owing to the connections, parts being simple bodies after decomposition, become elements.  From a certain point of view, composition of a whole out of parts is a reverse procedure with respect to decomposition of a whole into parts.  However, its product does not return the whole to its original state (Schedrovitsky 1964, 1965; Genisaretsky 1965).

The second group of procedures includes measurement of empirical objects and notation of their properties or aspects as formally distinct characteristics of all kind.  The “simple bodies” resulting from decomposition of an object can also be measured.  In this way, we will obtain both characteristics of the whole object and those of its parts.  Restoration of an object based on its characteristics is a procedure reverse to measurement.

The third group includes immersion of elements and their integrated structure into the whole, and the reverse operation of extraction of elements or structure out of the whole.

The aforementioned groups of procedures should be juxtaposed and combined in a consistent and non-contradictory system embodied in the elementary-structural ontological representation of a system object (similar to the way the arithmetic operations of addition-subtraction, multiplication-division, and exponentiation-extraction of a root are embodied in the natural number series). Unfortunately this has not been accomplished yet. As a result, a great number of methodological difficulties and paradoxes have emerged. Many of them are well known, the most typical being the paradox of identity-and-distinction of simple bodies and elements in chemistry, described long ago by D. I. Mendeleyev, and the paradox of material vs. nonmaterial nature of connections and structures.

When A. Lavoisier decomposed chemical substances into “elements” (in his understanding), and then combined a “compound” out of the “elements”, he interpreted his own procedures as a simulation in researcher’s actions of what is usually done by “nature.” Such an interpretation obligated one to ultimately describe everything that occurs or can happen in or to a system of chemical compounds, their entire “life”. This “naturalistic,” or “natural” description of system processes had to be derived from elementary-structural system representations based upon decomposition-composition of objects. However, this task was not accomplished, i.e., the task of establishing formal correspondences between processes occurring in a whole object and processes in its parts, or correspondences making it possible to look for material realization of certain processes or, which is a reverse procedure, to predict processes of the whole given the material organization of the parts. It was not accomplished either on the theoretical, or on the methodological levels in any single area of system research. This should not surprise us, because specification and description of natural processes of systems’ life were absolutely absent in the groups of procedures described above as underlying accepted and widely used ontological pictures and definitions of systems.  Nor are they to be found in many latest approaches to system analysis. This does not mean however that processes are not discussed.  Actually, they are always mentioned — as system “functioning”, as “system dynamics”, etc., but then most researchers reduce processes either to structural or to parametric characteristics.

Such a separation and even opposition of a “system” represented by its elementary structure, on one hand, and system processes occurring in the system, on the other, that is essentially equivalent to the statement that system is determined by its structure rather than by its processes, no longer corresponds to the practical and theoretical methods employed in many systemic areas of contemporary science and engineering.

To a large degree, the emergence of these new ways of work was due to the peculiarities of the design approach, but it later proliferated to science proper as well (Dubrovsky and L. Schedrovitsky 1971). The reason was that all classic natural sciences began their analysis with completely separate and physically distinct objects whose existence and laws of life, as postulated, did not depend on human actions. It was assumed that they were exactly as we found and perceived them. A complex system of research procedures (including measurements) was then applied to these objects which allowed researchers to discern, among other things, their inherent processes. Using special symbols, researchers represented these processes, thus separating them from the “material” of the objects and representing them as self-contained entities. Ultimately, they were interpreted as ideal objects existing separately and side by side with initially isolated physical objects. But no matter how these representations of processes were interpreted or viewed later, they always corresponded to physically distinct objects since they were obtained through their study.

Designers proceed in the opposite direction. Attention is centered on the product that a machine or a complex system must produce. Therefore, a designer must first determine the function of the object to which prescribed transformations are applied. This requires a knowledge and description of particular system functioning, while the material that could realize this functioning is of a secondary importance. Therefore, a designer does not start with a physically distinct object, but rather with ideally prescribed functions and functioning. Only then does he decide on the material that should realize these functions. In order to do this, he has to treat functioning as the object of his activity, and therefore, to represent it in the way that permits its composition, transmutation, and transformation, that are independent, to a certain degree, of its material. (This is so because material has to be selected later, depending on the functioning that has already been determined). What this means is that the designer starts specifying his object by determining processes that take place in it, first of all, the processes of functioning; and it is this process that defines the boundaries of the object of design as a system. Everything else has to fit them (Schedrovitsky 1973, 1969; Guschin et al.1969). This particular order of specifying and organizing the object that originally emerged in design, begins to infiltrate scientific disciplines that support design (of which many are gradually getting involved in design). Ultimately, this leads to a principle change in the type of the objects studied by modern science, along with structural transformation of scientific investigation itself  (Schedrovitsky 1970; Dubrovsky and L. Schedrovitsky 1971).

Taking into account all these processes, we must conclude that modern system approach, actually existing and developing in design and in new scientific disciplines, can no longer be based on the traditional concept of system, that supplants all the above-mentioned procedures of parametric measurement, decomposition-composition, and immersion of parts into the whole. System approach can no longer be satisfied either by the ontological representation of a system as a set of elements united by structure, or by the concepts of “element,” “relationship,” “dependence,” “structure,” etc., serving this ontological picture. Modern system approach requires a totally different procedural basis, and therefore, a different ontological picture of a “system” that envisages different aspects of it as an object and subject-matter of study, and in different correspondences, too. Accordingly, the main categories of system approach and of the concept of system itself will change.

The essence of the new approach can be summarized in a very simple principle. To treat any object as a complex system means sequentially to represent it in four categorical planes of specificprocesses, functional structure, material arrangements, and morphology, and then to “unfold” morphology again on the above four planes; and repeat this procedure until an adequate concrete representation of the object is obtained. Graphically, the content of this principle can be represented by the Figure: every unfolding of the scheme in a column constitutes a step of system analysis that represents the object as a simple system. Owing to the reverse procedure of collapsing the second system representation into the “morphology” of the first system, morphology appears as a special layer in a simple system representation that, relative to itself, organizes the three other planes into the second layer.

The new concept of system has several advantages compared with the earlier one. One of them is the possibility to easily connect any process representation, including evolutionary-genetic, with structural and organizational representations. Another advantage is that the problem of interaction among systems can be solved without any difficulties and paradoxes. While previously, any presupposition of interaction of systems immediately transformed it into elements of system of interactions, now systems can interact on the level of material without affecting either their wholeness, or autonomy of their functional structures and processes. We believe that even these few considerations are sufficient to justify the conclusion that the new concept of system is an effective one having a great potential, provided that effort and time is devoted to its further development.

* * *

This paper was written by Georgy Petrovich Schedrovitsky (1929-1994), who directed Moscow Seminar on General Methodology (also known as Moscow Circle on Logic) from the late 1950’s till his death in 1994. Although it was published in Russian (Shchedrovitsky, 1974), I believe that its publication in English is timely and necessary. Special thanks to Dr. Emanuel Smikun for his help in translating this article. The translation and the abstract were prepared by Vitaly Dubrovsky.


  1. Genisaretsky, O.I. (1965). “Characteristics of Objects of System Research.,” in Problems of Research of Systems and Structures (Problemy Issledovaniya System ee Structur)., Moscow, Russia.
  2. Gushchin U. F., Dubrovsky, V. J., and Shchedrovitsky, L. P. (1969).  ”A Concept of System Design,” in Large Scale Information‑Control Systems (Bol’shiye Informatzionniye Upravlyayooshchiye Systemy )., Moscow, Russia: Znaniye.
  3. Dubrovsky, V.J. and Shchedrovitsky, L.P. (1971.) System Approach to Human Factors Engineering (Systemnoye Ingenerno-Psychologicheskoye Proyektirovaniye), Moscow UniversityPress, Moscow, Russia.
  4. Shchedrovitsky, G.P. (1964). Methodological Problems of System Research, Moscow, Russia. (English translation in General Systems, 11, 1966). Reprinted in Selected Works of G.P. Shchedrovitsky (A.A. Piskoppel and L.P. Shchedrovitsky, eds)., Moscow, Russia, 1995.
  5. Shchedrovitsky, G.P. (1965). “On Principles of Classification of the Most Abstract Directions in Methodology of Structured-System Research,” in Problems of Research of Systems and Structures (Problemy Issledovaniya System ee Structur)., Moscow, Russia.
  6. .Shchedrovitsky, G.P. (1969). “The Problem of Object in System Design,” in Proceedings of the Second All Union Conference on Technical Cybernetics., Moscow, Russia.
  7. Shchedrovitsky, G.P. (1970). “On System of Pedagogical Research: Methodological Analysis,” In Optimization of Education Processes in Higher and Secondary Education (Optimizatziya Processov Obucheniya v Vysshey ee Sredney Shkole). Dushanbe.
  8. Shchedrovitsky, G.P and Dubrovsky, V.J. (1973). “The Problem of Object in System Design,” in Methodology of the Design Activity Research (Metodologiya Issledovaniya Proektnoy Deyatel’nosty), Vol 2., Moscow, Russia.
  9. Shchedrovitsky, G.P. (1974). “Two concepts of system,” in Proceedings of the XIII International Congress on History of Science and Technology., Moscow, Russia. Reprinted in Selected Works of G.P. Shchedrovitsky (A.A. Piskoppel and L.P. Shchedrovitsky, eds)., Moscow, Russia, 1995.

In J. Willby, ed. Proceedings of the 46th Annual Conference of the International Society for the Systems Sciences, Asilomar, CA, July, 2002, 5 p.