Simon – The Architecture of Complexity

Simon, H.A. . The Architecture of Complexity . Proceedings of the American Philosophical Society, Vol 106, No 6, pp. 467 – 482 . 1962

Development of a general systems theory to find out which abstracting properties from all of them can apply to all kinds of systems. Do diverse systems have anything non-trivial in common? This is addressed by ideas under the umbrella of cybernetics (if not a theory than at least an interesting point of view). The goal is to cast some light on the ways complexity exhibits itself wherever it is found in nature. The rough description of a complex system used here is: a system made up of many parts which interact in a non-simple way. In such systems the whole may be more than the parts in a pragmatic sense: ‘In the face of complexity, an in-principle reductionist may at the same time be a pragmatic holist’ [Simon 1962 p 468]. How complexity frequently takes the form of hierarchy is discussed in four sections: 1) frequency of the occurrence of hierarchy in complex systems 2) hierarchic systems evolve more quickly than non-hierachic systems 3) dynamic properties of complex systems and they can be decomposed into subsystems 4) relation between complex systems and their descriptions.

>HIERARCHIC SYSTEMS

A hierarchic system is a system that is composed of other interrelated hierarchic systems. DPB: a hierarchic system integrates other hierarchic systems until some lower, elementary level of subsystems is arrived at. What that level is, is somewhat arbitrary and how it can be done is a subject of this article. Hierarchy is often referred to the structure where systems are subordinated by a relation of authority to the system they belong to. This means the existence of a boss and subordinate subsystems. Each system has a boss who is subordinated to the boss of the system. This is a formal approach to hierarchy. ‘I shall use hierarchy in the broader sense introduced in the previous paragraphs, to refer to all complex systems analyzable into successive sets of subsystems, and speak of ‘formal hierarchy’ when I want to refer to the more specialized concept’ [Simon 1962 p 468].

>>Social Systems

One kind of hierarchy in social sciences is the formal organization of businesses &c. Another is families, tribes, clans, &c.

> >Biological and physical systems

Cell-up Cell>tissue>organ>system. Cell-down: Cell>nucleus>mitochondria>membrane>microsomes.

Elementary particles, Planetary systems

A gas is seen as a random distribution of complex systems, namely particles.

Hierarchy refers to a system with a moderate number of subsystems with their subsystems (a diamond is a flat hierarchy with many subsystems, and atypical). The number of subsystems subordinated to the system is the span of that system. If the span of a (sub) system is wide it is flat at that location. A diamond has a wide span / is flat at the crystal level, but not at the molecular level. Biological and physical systems differ from social systems in that the first are described in spatial terms and the second by defining who interacts with whom. This can be reconciled by defining hierarchy by intensity of interactions.

>>Symbolic systems: Books>Chapters>Paragraphs>Alinea>Words>Letters, &c.

>THE EVOLUTION OF COMPLEX SYSTEMS

Watch maker 1: One system. When assemblage is interrupted the entire watch falls apart. Watch maker 2: Subsystems of 10 subsystems each. When assemblage is interrupted the subsystem at hand falls apart. This one is more likely to survive.

>>Biological evolution

The time required for the evolution of a complex form from simple elements depends critically on the numbers and distribution of potential intermediate stable forms’ [Simon 1962 p 471]. Comments: a) no teleology is suggested and the structure can come from random processes. When complex forms once existent become stable they give direction. But this is survival of the fittest, namely survival of the stable b) not all large systems appear hierarchical c) the evolution of complex systems from simple elements implies nothing concerning the change of entropy: free energy can be taken up or generated by the evolutionary process

>>Problem solving as natural selection

Problem solving requires selective trial-and-error. .. In problem solving, a partial result that represents recognizable progress toward the goal plays the role of a stable sub-assembly’ [Simon 1962 p 472]. Human problem solving involves only trial-and-error and selectivity. The selectivity derives from heuristics to suggest which paths to try first.

>> The sources of selectivity

When we examine the sources from which the problem-solving system, or the evolving system, as the case may be, derives its selectivity, we discover that selectivity can always be equated with some kind of feedback of information from the environment’ [Simon 1962 p 473]. DPB: the approach to modeling evolution is the same as that to modeling problem solving. There are two paths of selection in problem solving: a) various paths are tried out, the results are noted and this information is used for further search and b) using previous experience: doing the same paths that lead to an earlier solution. In this way trial-and-error is reduced or eliminated. The closest analogue of this in organic evolution is reproduction.

>>On empires and empire building

When an empire breaks up, it doesn’t tend to fall apart into its smallest elements but into the next scale of subsystems.

>>Conclusion: the evolutionary explanation of hierarchy

Systems will evolve from stable intermediate forms faster than from basic elements to form hierarchies, the subsystems based on the intermediate forms. Hierarchies have the time to evolve.

>NEARLY DECOMPOSABLE SYSTEMS

A distinction can be made between the interactions within subsystems and between them. Their intensity and their frequency is different to orders of magnitude. Employees within the formal organization of a department have more and more intensive contacts than employees of different departments. The decomposable case can be used as a limit over a wide range. In the nearly decomposable case the interactions between the subsystems are weak but not negligible. From the latter case these can be proposed: a) the short run behavior of the subsystem is independent of that of the other subsystems and b) in the long run the behavior of a subsystem depends on the behavior of the others in the aggregate. This is illustrated with an insulated house within which there are somewhat insulated rooms within which there are hardly insulated cubicles. A change of the temperature in the rooms, induces a rapid change of the temperature between the cubicles, but a slow change of temperature between the rooms. If a complex system can be described with a nearly decomposable matrix then the system has the properties a) and b) above.

>>Near decomposability of social systems

Most of the communication channels in formal organizations are between employees and a very limited number of other employees. The departmental boundaries are assumed to assume the same role as the walls in the thermal insulation example.

>>Physico-Chemical examples

The theory of the thermodynamics of irreversible processes, for example, requires the assumption of macroscopic disequilibrium and microscopic equilibrium, exactly the situation described in our heat-exchange example’ [p 476]. DPB: how does this work?

>>Some observations on hierarchic span

Suppose that the elements of a system have properties for stronger bonds and for weaker bonds and that the stronger bonds exhaust through the bonding. Subsystems form through the strong bonds until they are exhausted. Then the subsystems will be linked by the weaker second-order bonds into larger systems. In social systems the number of interactions is limited by the serial character of human communication (one at the time) and limitation on the time consumption involved in a role and hence of the number of roles one can handle (one can have a group of friends consisting of a dozen but not hundreds).

>>Summary: Hierarchies tend to be near-decomposable.

>THE DESCRIPTION OF COMPLEXITY

People draw complex objects in a hierarchical way. The information about the object is arranged hierarchically in memory, like a topical outline. DPB: re active association. When information is presented in this way the relations between a subpart and another subpart can be presented and between subsubparts within each. Information about reations between subsubparts of different subparts is lost.

>>Near decomposability and comprehensibility

By representing parts hierarchically little information is lost (re b, aggregate effect above). Many complex systems have a near-decomposable, hierarchical structure. That enables us to see them. If complex systems exist that are not so structured then they are unobserved and not understood. ‘I shall not try to settle which is chicken and which is egg: whether we are able to understand the world because it is hierarchic, or whether it appears hierarchic because those aspects which are not elude our understanding and observation’ [Simon p 478]. DPB: the processes that brought forth our powers of perception and the processes in nature are fundamentally the same.

>>Simple descriptions of complex systems

There is no conservation law that prescribes that the description of a complex system should be as complex as the system itself. Example of how such a system can be described economically, or, in other words, how it can be reduced. This is only possible if there are redundancies in the system. If it is completely unredundant then the system is its own simplest (most economical) description and it cannot be reduced. DPB: this notion of reduction is exactly the opposite of the notion used by Ashby. He uses reduction to indicate the opposite of organization. That which is not organized can be reduced (away) until organization remains. This is the mathematical definition of reduction. Here it is the opposite: whatever is redundant leaves room, or in other words can be reduced to, a rule. Three forms of redundancy are: a) hierarchic system is often assembled from a few kinds of different subsystems in various arrangements. DPB: this is a form of repetition of the components used. b) ‘Hierarchic systems are often nearly decomposable. Hence only aggregative properties of their parts enter into the description of the interaction of those parts. Not the lower level properties of the composing elements of the parts play a part in the interactions of the components at the higher levels. A generalization of the notion of near composability might be called the ‘empty world hypothesis’: most things are only weakly connected with most other things’ [Simon 478]. DPB: This means that some properties of the subcomponents of a complex hierarchical system, which are themselves built of subcomponents, enable interaction with other subcomponents, and form the complex hierarchical system that they are a subcomponent of. But those enabling properties of the subcomponents are properties of, or based on properties of the subcomponents of the subcomponents that form the complex hierarchical system. ‘The children are not allowed to participate in the discussion between the family elders’. Given that emptiness can be described by the absence of a description it can be described economically. DPB: how is this a form of repetition? The aggregative properties of the subcomponents of the system repeat, and they are based on repeating or comparable properties of the sub-sbcomponents. c) Redundancy can originate in a constant relation between the state of a system and a later state of it. DPB: this is a form of repetition of the behavior of the system. It can be a literal repetition or the lingering of a system like a kind of an after image of the previous state. In any case the current state can be compared with the previous one and with the next state also. Cognition is the application of its powers to compare, identification of redundancy, so as to perform a (re)cognition of recurrence of coherence, namely pattern. On a devised continuum of cognition ‘a suspicion’ is on the one extreme, where a pattern merely reminds of something such that it cannot be predicted for what it ‘is’ or whether it will occur with any certainty. ‘Knowing’ is at the other extreme, where the pattern is known and its occurrence can be predicted with a high level of certainty.

>>State description and process description

State -: a circle is an object of which all points are equidistant from one point. Process -: hold one arm of the compass in place, rotate the other arm until is is back at the initial point. ‘These two modes of experience are the warp and weft of our experience. .. The former characterize the world as sensed; they provide the criteria for identifying objects, often by modeling the objects themselves. The latter characterize the world as acted upon; they provide the means for producing or generating objects having the desired characteristics. The distinction between the world as sensed and the world as acted upon defines the basic condition for the survival of adaptive organisms. The organism must develop correlations between goals in the sensed world and actions in the world of process. When they are made conscious and verbalized, these correlations correspond to what we usually call means-end analysis. Given a desired state of affairs and an existing state of affairs, the task of an adaptive organism is to find the difference between these two states, and then to find the correlating process that will erase the difference. Thus, problem solving requires continual translation between the state and process descriptions of the same complex reality’ [Simon 1962 p 479]. DPB: this is my equalizing of differences. It refers to adaptive organisms that is autopoietic systems. The translation between state and the process are then the same as the recurring consequence of structure and operations: the description of what it is and the description of what it does, &c. Refer to this in the main theory. ‘We pose a problem by giving the state description of the solution. The task is to discover a sequence of processes that will produce the goal state from an initial state. Translation from the process description to the state description enables us to recognize when we have succeeded’ [Simon 1962 p 479]. DPB: can this be coupled to challenge propagation?

>> Ontogeny recapitulates phylogeny

If genetic material is seen as a program, it: a) is self-reproducing, b) developed by Darwinian evolution. A human develops gills and then use them for other purposes. Instruct a 20th century workman to build a car by what he knows: start with a cart, remove the singletree then build a motor onto it, then a transmission. &c. DPB: this does not necessarily apply to the memetic instructionset of a firm. Or is it, sometimes routines are in place that stem from previous versions of work instructions that are no longer in place: ‘The generalization that in evolving systems whose descriptions are stored in a process language, we might expect ontogeny partially to recapitulate phylogeny has applications outside the realm of biology. It can be applied as readily, for example, to the transmission of knowledge in the educational process. In most subjects, particularly in the rapidly advancing sciences, the progress from elementary to advanced courses is to a considerable extent a progress through the conceptual history of the science itself. Fortunately, the recapitulation is seldom literal – any more than it is in the biological case. .. But curriculum revisions that rid us of the accumulations of the past are infrequent and painful’ [Simon 1962 p 481]. DPB: this is an important thought concerning the execution, namely the enactment of memes ad how the are restricted by the actual state of affairs, when the firm is operational.