A realist critique of the use of system in historiography

By Haines Brown, November 2006

The idea of “system” has appealed to historians, particularly those who are scientifically inclined, because it seems to promise a way to bring together a multiplicity of causal factors into a meaningful whole.

This hope is a worthy one, but is seldom realized because it does not look beyond a description of surface appearances to grasp the inner mechanisms at work—why the system parts exist, the nature of their relation, and why that relation gives gives rise to the phenomenon associated with the whole.

First, let me start with a conventional definition. There seems to be agreement that a system is an aggregate of parts that are causally connected, and the result of their interaction is the emergence of behaviors or properties of the whole not found in its parts. I will suggest this definition is inadequate because it is phenomenological.

The non-reductionism implied here was undoubtedly another reason for the appeal of a system approach, for it seemed to be scientific without at the same time embracing the positivisitism, mechanical determnism, or atomism long popularly associated with the natural sciences.

Years ago there was a debate over “wholes and parts” that today seems rather quaint. The reductionist side argued then that a whole can be explained in terms of the qualities and behaviors of its parts, but it was all too obvious that this is not the case in the real world, and no one would defend the position today. However, the issue was not really settled, for if the system in not reducible, its parts must in some way (in non-equilibrium systems) contain or represent the system as a whole within themselves, and it has proven very difficult to pin down this quality of being part of the whole. Unfortunately, the efforts have tended to remain highly intuitive or speculative. They do not really come to terms with the material basis for the system dimension that is necessarily present in all non-equilibrium systems, although it cannot be observed empirically.

In a salt crystal, the cubical crystalization, white color, and other features are observable evidence of a wholeness quite distinct from the characteristics of sodium and chlorine atoms. A static or equilibrium system manifests features of the whole not found in its parts, which is why we call it a system.

But this should not be carried over to the non-equilibrium systems of interest to historians, which are not only manifest in terms of system behavior rather than empirical qualities, but also the parts acquire an unobservable system dimension. Unlike the equilibrium system, the quality of being non-eqilibrium is not empirically manifest, but arises from the inner mechanism of the interaction of parts that are at the same time both distinguishable parts and manifestations of the whole.

For example, the homeostatic behavior of the earth's climate up to now is not explained by taking all relevant factors into consideration and predicting the likely outcome of their causal relations, but by exposing the structure and function of those parts in the whole that depends on features specific to both the parts themselves and specific to the whole. A negative feedback means that the contribution of a part to the whole is reduced because of the state of the whole; the part changes because of the whole and the whole changes because of the parts.

In human society, if we fail to see its parts as being systemic in nature (seeing humans a social beings), we are forced to adopt an empiricist reductionism, either embracing an unrealistic social atomism or an undesirable mechanical determinism, where society is either the manifestation of human nature or humans are entirely determined by the society of which they are passive members.

For an example of the application of the system idea in history, World Systems Theory takes a trade cycle as evidence of the systemic nature of world commerce. But is it? There are cyclic phenomena that don't manifest a system. For example, the rotating shadow of a sundial is cyclic, but not because the sundial and the shadow form a system. We know the behavior is really due to the behavior of another system, the rotation of the earth in relation to the sun, and we know this is the case because we can find no causal mechanism in the sundial that would account for the the rotating shadow. Without looking for a system's inner causal mechanisms, one remains at the phenomenal level, which can cause a variety of difficulties.

Another problem is that World Systems Theory's phenomenalist approach fails to distinguish a world system of trade from ordinary trade connections. Both are a source of profit (buying where goods are abundant and selling dearly where they are not will do that), and volume and profit levels will naturally fluctuate in either case. And yet what World Systems Theory wants to do is to distinguish a World System from other large-scale systems (and not with a claim that “World Systems” are necessarily global in extent).

Such a distinction requires that the participants in a world system of trade in some way incorporate that systemic quality to distinguish it from ordinary trade networks that are not “World Systems”. This system quality embedded in its parts is clearly not observable in a reductionist view of things, but instead is unobservable. We could do one of two things about these unobservable features: treat the unobservable systemness as being real and seek the causal mechanism that links the whole with its parts within its parts, or what is usually done is treat the unobservable as merely a useful artiface that associates the observed phenenoma of trade participants and the trade cycle without trying to explain it.

A result of the latter approach is the marginalization of the reason for trade profits in the first place, for their mere existence suffices. In early classical political economy, it seemed that trade brought wealth to nations because it was associated with a division of labor, which was felt to imply increased productivity. In other words the source of wealth lay not in the “sphere of circulation” (trade), but in the interdependency of two systems, the sphere of production where wealth is produced and the sphere of circulation where the wealth is realized in observable form. To focus on trade cycles has the effect of a substitution of profitability for the generation of the wealth from which profits arise. Of course, the sphere of circulation will be subject to fluctuation such as due to changes in supply and demand, but that behavior really says very little of significance about anything outside commerce. It is only when the systems of production and exchange are brought together into one larger system that we begin to come to terms with the modern economy. Otherwise one is seduced by the fluctuation of profits to project the world system into all times and places. The modern (capitalist) economy thereby gains a specious universality that has serious ideological implications. Without exposing the inner causal mechanism and by remaining at the phenmenal level, this absurdity easily results.

Another problem is the subjectivity present in most employments of system features in historical analysis. The proponants of a system view have the admirable goal of being open to a wide range of different kinds of factors that in some way can be viewed as interdependent because they participate in a system. These factors might otherwise not lend themselves to simple causal explanations that require events that are of a comparable nature and proximate in time and place. A systems approach would seem to escape these limitations of the conventional notion of causality, but unfortunately it often ends by mystifying the situation. Part of the reason is the presence of a subjective factor that is usually unperceived.

If wholes display behaviors or characteristics that are not found in its parts, it might appear a simple matter to use those “emergent” qualities of the whole to define and identify the boundary of the system. The boundary is the point at which those emergent qualities cease to manifest themselves. Unfortunately, this is not so easy. It might be assumed that a boundary (whether ideal or material) represents the extent of the characteristic systemic behavior and distinguishes the system from the surrounding world, but boundaries are not as simple as placing eggs in a basket.

There are cases in which

In the science-fiction Star Trek series, the crew of the Enterprise encounters alien beings who insist upon characterizing humans as “ugly bags of water”. We would not be inclined to describe ourselves that way, but the aliens were not being inaccurate. It all depends on one's perspective. Today the economic system dominates our lives, but under feudal conditions, it did not in the sense that most production was consumed by the producer and the community of producers and aimed at subsistence, not profit. To project a perspective based on the modern economy into the distant past is to impose values that are surely alien to it.

So, then, are boundaries and system wholeness just arbitrary and subjective? Some appear inclined to admit it. However, there is an escape from subjectivism when the behavior of a whole dictates the existence of a boundary. It turns out that there are two opposite kinds of system behaviors that are interdependent and therefore do imply a boundary between them.

First, we know that a hypothetical temporarily isolated system necessarily experiences an increase in its entropy, which is generally called “dissipation”. This word “entropy” is rather ambivalent and hard to define, but it is real enough and employed in many areas of study with success. When a system becomes less ordered, has less capacity to bring about change, and moves toward a more probable state, its entropy is said to increase. This behavior seems a consequence of how the system came about in the first place, which is not dependent on a boundary. While entropy is usually calculated for ideal isolated systems (in order to define their state in empirical terms), the fact that a real system is open does not mean that the law ceases to have effect (a point is broadly defended in the sciences), but only that we have trouble assigning a specific value to it.

However, there is no question that some sub-systems move in the opposite direction and actually decrease their entropy. This behavior, called “emergence”, is characteristic of human history. It is achieved by the existence of a mediating boundary between two sub-systems that constrains their causal relation such that the increase in entropy of one-system acts as an “engine” to decrease the entopy in the other subsystem, in effect exporting its entropy. The boundary must be material, and the net effect of these two opposite behaviors must be a increase in entropy.

So in the case of the emergent systems that characterize human affairs, it is possible to define a boundary. One can objectively specificy what is emergent because it is empirically novel; one can objectively specificy the other subsystem to which the first necessarily exports dissipation because we can observe its effects; and we can identify a material mediation of the relation of these two subsystems that explains the interdependence of their opposite behaviors. Only when dealing with systems being improbably emergent can we define a real boundary, which is the mediation between them. Otherwise, boundaries are artificial constructs we impose on reality.

Entropy is another system quality that can't be observed, although very real. It is not my aim here to explore its implications beyond suggesting that its use is the only way to define a system boundary objectively. For example, the forces of production are a mediation (boundary) between society and nature such that the dissipation of nature is an engine to create new economic value in production, we have used an unobservable inner mechanism to arrive at an explanation.

It seems that the causal relation of system parts—the structure of the system—is not only an important aspect of a system, but, to avoid a mechanical reductionism, that structural quality must be represented as a real aspect of the system's parts even though not observable. A view that is gaining ground in the philosophy of science is that unobservables (process, causal relation, entropy, etc.) are indeed real. People who hold to this position are termed “scientific realists”, and they believe that it makes possible a theoretical reductionism that integrates the theories specific to different domains of reality by seeing them as themselves determined by the underlying nature of things. For example, while human history does not reduce to the laws of physics, both physics and society are emergent features of matter.

The opposite and more traditional view can be described as instrumentalist and empiricist. In this case, unobservables are theoretical constructs associated with regular behaviors which, being empirical and regular, are subject to empirical verification. However, they are not real and therefore can't be reduced to achieve a real explanation of behavioral rules or an integration of our knowledge of the world.

While it might appear these two positions do not really differ much in practice, such is not the case. Human history is an emergent process in which regularities or laws play only a minor role. Civilizations don't rise and fall because there's a law out there that they must do so, but they rise and fall for specific reasons that underly specific explanations. To look for regularities as evidence of the existence of a a system in human affairs seems to downplay the creative or emergent character of human society, of what makes us human.

A systems approach naturally appeals to historians, but it has been employed with little real success. The concept remains undeveloped because historians often assume a Cartesian mind/body dualism and a radical empiricism that are incompatible with a systems approach that is realist and escapes subjectivism.