|Time Warp--On Science--On Health-- On Defense-- Artcam|
The intelligent design of complex systems requires an understanding of the relevant inter-relationships of the parts to the whole. In so far as our understanding is faulty or less than complete, we are left open to unforeseen interactions that can cause catastrophic failure of the overall system. As designers and analysts, at every level of description we strive to separate the chaff from the wheat and to recognize the difference between the chaff and the grain. In a way this process is a metaphor for how we should approach our lives. (AW)
This article was originally published in 1969 and is reprinted here from the book Biomechanics by permission of Plenum Publishing Corporation. When we wrote the article, John was professor of Engineering and Psychology at UCLA and I was one of his graduate students. This article is an outgrowth of the many hours we spent discussing how to approach the design and analysis of complex systems and how to avoid catastrophic system failures arising from not considering critical sub-system interactions. Unfortunately, since this article was published many examples of avoidable system failures can be cited: among these, the Challenger explosion; the 1999 failed Mars mission; the last presidential election where deficiencies in voting machines and questionable ballot "design" quite possibly affected the ultimate outcome of the election; and recently, the Columbia tragedy where the absence of an on-board viewing device made the extent of lift-off and/or in-space injury to the spacecraft difficult to ascertain.
John had a long and fruitful career. He was a teacher, a futurist, a researcher, an expert in prosthetics and an expert in human factors engineering. He never stood on formality; he was on a first-name basis with most of his graduate students and gave them their head. He stimulated their creativity. He was a true interdisciplinary thinker - and most of the students who gravitated to his Biotechnology Laboratory were attracted to that vision. His frequent Friday night parties were legend (some would say, notorious) and no one who knew him could forget the heavily spiked but innocent-tasting punch he used to serve his initially-unsuspecting victims.
John passed away in May 2001 after a relatively short illness. Over the years John and I became close friends and his death leaves a void in my life that cannot be filled. Needless to say, he will be sorely missed by his loved ones and by his many colleagues, students and friends. I am reprinting this article in his memory because he felt the article wore well and because the concepts we wrote about then are are as relevant today as they were when we wrote them. - Gary Fisher
In the tradition of the well-known legend about the six blind men examining an elephant, the performance of both living systems and man-made systems can also be described from many points of view. Figure 1 shows some representative levels of description. Descriptive levels are a convenient construct which aid in organizing knowledge and understanding both complex living and nonliving systems. They are observer-generated: that is, the perception of what a unit is at a level depends on the ability of the observer to discriminate differences at that level and between that level and other descriptive levels. The discovery of these levels is intimately connected with our measurement process and with extensions of our senses, e.g., the microscope, which aid that process. When these discovered levels are ordered in a hierarchal fashion, e.g., by size, the discriminable units at each level become somewhat analogous to structural building blocks. For instance, molecules are a structure of atoms, cells are a structure of molecules, tissues are a structure of cells, and so forth. This "structural peeling of the onion" is interesting in itself; however, what one is generally interested in is not the structural breakdown of the "onion," per se, but rather the manner in which structure and function are integrated at and between the various descriptive levels.
When examining a particular system, whether living or man-made, the choice of a specific level of description is important because it may reveal interlocking relationships with other levels that can critically affect the overall performance and/or survival of the total system. Thus, for example, a few molecules of LSD may produce profound behavioral changes in the total organism.
It is the thesis of this chapter that a systematic examination among the levels of description will reveal relationships useful in the design of man-made constructions. Also, from the standpoint of understanding more about life systems for their own sake, such an examination might yield critical insights into the processes by which a given life system survives. We would like to develop a few examples to demonstrate this view.
Our first example is from Decker (1959), who, in his studies of forest succession, visualized the life history of a tree as a "plane of events" which could be plotted graphically as a series of ordered size sequences varying with time on a horizontal axis. With this approach one can choose several descriptive modes; for instance, the choice might be in terms of events concerning the seed, sapling or the tree. Alternatively, the same life history could be stated in terms of chains of organ events concerning what happens to leaves, roots, and stems or as events occurring at the cellular, subcellular, colloidal, and molecular structures. Application of the plane of events concept to various successional sequences provides a visual means of interrelating events over time at the various levels of description we have shown in Fig. 1.
Decker illustrates the importance of choice of a particular level of description as follows:
"During a graduate seminar a question arose as to why eastern white pine (Finus strobus L.) does not occur on certain good hardwood sites in the Adirondack Mountains. A lively and speculative discussion developed that centered on comparative chemical properties of soils on these sites and on nearby sites where white pine grew well. Upon checking the problem and the discussion against a plane of events, the following became apparent: the discussion was based on an implicit assumption that the problem event behavior of whole plants) was a consequence of certain prior molecular events (chemical properties of soil); none of the intervening events had been established or even identified; the unrecognized question of prime importance was at the whole plant level and concerned the stage at which the life cycle of white pine was interrupted on these sites; chemical properties of soil might have no important relationship to the interruption; the premature discussion at a molecular level was a kind of inferential broad-jump across the plane of events. (p. 155)
Environmental factors such as shading from sunlight at a critical stage of development could well be the necessary condition for life continuance and, accordingly, must be explicitly identified as part of the description.
The great degree to which structure and function is integrated \vithin and between levels in living systems is elegantly illustrated by the process of wound healing (Weiss, 1956). Figure 2 illustrates in diagrammatic cross section the consecutive steps in the repair of a skin wound. The wound (B) is rapidly covered by an exudate (C). This is followed by detachment from the basement lamella of the epidermal cells bordering the wound and mobilization of these cells to finally cover the exudate (D). The process of formation of the missing basement lamella between the epidermal coat and the raw tissue beneath (D-E) illustrates an orderly progression (or hierarchal structuring), through time, from lower to higher levels of organization. We will briefly outline this process here.
The intact basement lamella is composed of about twenty layers of cylindrical fibers, the direction of which changes from layer to layer at approximately right angles. Prior to lesion the epidermal cells are glued to the basement lamella by a single intermediary layer of evenly spaced granules about 600 Å thick. Several days after wounding the intermediary granulated layer is restored and underneath the exudate a group of fibroblasts appears. According to Weiss, the bulk of evidence indicates that the elements for the fibers that ultimately form the basement lamella come from these cells.
During the second week after wounding the redeveloping basement lamella consists of a matrix of young fibers in random disarray. This matrix bears little resemblance to the ordered structure of the intact basement lamella. The important point is that the building blocks necessary for the reconstruction of the basement lamella are demonstrably present at the building site, not as macromolecules but as already-formed fibers, before there is any sign of a higher-order organization of the basement substrata.
The next stage in the process is the orientation of the randomly arrayed fibers into orthogonal layers. This begins at the epidermal side with a layer of fibers running parallel to the surface. As soon as layering appears, the alternative layers assume the orthogonal orientation typical of the intact basement lamella. Weiss (1956) summarizes that the reorganization of the fibers "definitely starts out as a planar pattern, which then extends itself into a space lattice as geometric order sweeps from the epidermal surface downward to align the erstwhile random population of fibrous elements into the characteristic layered grid this is clearly a case of organization of a higher order emerging stepwise through the integration and regrouping of lower-order units." (p.825)
This emergence of higher-order organization resulting from step-by-step integrations of lower-order units is not confined to the regrowth process just described but can be observed in a wide range of life systems. In the building of the T-4 virus, for example, the construction "seems to follow an assemblyline process with the three major branches that lead independently to the formation of heads, tails, and tail fibers. The finished components are combined in subsequent steps to form the virus particle." (Wood and Edgar, 1967, p.63). The slime mold amoebas, as still another example, gather to form a stock culminating in a mass of spores which, when they are dispersed, can each split open to liberate a new amoeba (Bonnder, 1959). This process exhibits a step-by-step integrative behavior analogous to that discussed in the previous examples.
In growth or regrowth processes such as we have illustrated it is evident that each step-by-step integration provides the environmental interface necessary to proceed to the next higher component level; the environmental configuration on the level of the total system is thus a highly complex combination of the environments of subunits at the intervening levels.
Turning now to certain features of the design process in an engineered system, we can see how achievement of the intimate interlevel structure-function relationship is, by analogy, the very essence of success or failure of the system. Figure 3 illustrates the levels which must be considered for any design and especially for designs with complex organizations of man and machines. The mission and the system-complex levels are initially formulated in a more or less philosophic manner in the designer's mind, based on a need, State-of-the-art and economic factors and other physical, political, and otherwise pragmatic constraints determine the feasibility of realizing the other levels implicit in the mission. The initial engineering design activity thus usually follows a pattern that starts from the mission specifications and then proceeds in steps by identifying components down through the various levels. In other words, the basic design of man-made systems is directed from the outside to the inside.
The man-made process is generally in contrast to the evolutionary development in living systems, which starts from random interactions on the materials level. The mission configuration in living systems develops upward with the relevant survival environment becoming successively more complex. This is design from the inside to the outside. Thus, performance on a gross system level is not specified at the outset of the evolutionary process, but results from an integration of structure-function from each level up, subject to the imposed environmental constraints at each level. The complex vertical and horizontal structure-function relationships existing in living systems is the logically inevitable outcome of environmental constraints occurring at various stages in development. The continued existence of the system depends on the successive restructuring of interlocking relationships at the microlevels which favor the survival of the system as a whole.
This contrast between man-made design and the evolution of living systems is not, of course, absolute; man-made designs go through numerous iterations during the development, testing, and operational life cycle. However, there are large gaps among the relevant levels of function, gaps which place man-made designs at a serious disadvantage in direct comparison. For example, the process by which a self-sealing fuel tank repairs itself can hardly be considered analogous to the wound repair described above. Yet, in some cases, as with the laser and the transistor, the gap between levels may be enormous and still provide valuable functional properties. In these latter two examples understanding of events at a microlevel has allowed the realization of missions which would otherwise be left out of practical consideration - and even, perhaps, human conception. One can only speculate about the possibilities that could arise in complex engineered designs if more attention were given to the explicit understanding of level interfaces at a more detailed and intimate degree such as may be under way in some of the current microcircuit technologies. Such an approach is certainly an important aspect of a design philosophy and should ultimately provide added human-valued utility, to say nothing of reliability. For example, would the Apollo tragedy of early 1967 have occurred if greater attention had been given to the Level 7 (life support materials) and Level 4 (escape subsystem) interface? Hind-sight is rarely a benign teacher.
The notion of attention to detail is certainly not new and as measurement technology at all levels has improved, so has sophistication in its application. Moreover, even the design philosophy concerning cross-level interface problems has been well articulated (1964 NASA Authorization). However, as a practical matter there is much to be learned concerning the actual complexity of a system under design and development and also the methods for identifying regions where there are technological unknowns. A simple index of complexity is a total count of the interfaces that can be recognized. By preparing a matrix of these interfaces for more detailed examination subdivisions can be examined and assessed for their significance on a scale of costs and values. In addition, areas where there is lack of technical knowledge can be determined, and the extent of the gaps within and between the levels of description can thereby be evaluated. Though such procedures may require extensive human and computer support, they assure that at least first-order criteria can be established for estimating the survival probability of complex man-made designs. If one doubts the payoff implied, he has only to look at the elegant natural examples surrounding him.
Bonnder, J. T., "Differentiation in social amoebae," Sci. Am., 201:152-162 (1959).
Decker, J. P., "A system for analysis of forest succession," Forest Science, 5:154-156 (1959).
NASA Authorization, 1964, Hearings before the Subcommittee on Space Sciences and Advanced Research and Technology of the Committee on Science and Astronautics, U.S. House of Representatives, U. S. Government Printing Office, Washington, D.C., 1963.
Weiss, P. A., "The compounding of complex macromolecular and cellular units into tissue fabrics," Symposium on Biomolecular Organization and Life Processes. Proc, Nat. Acad. of Sci., 42:819-830 (1956).
Wood, W. B., and Edgar, R. S., "Building a bacterial virus," Sci. Am., 217:60-75 (1967).
© Copyright 1969 Plenum Press
Lyman, John and Gary Fisher "What's Inside the Outside." Biomechanics, D. Bootzin and H.C. Muffley, ed., Plenum Press, New York, p. 75-80 (1969)
Reprinted on this Website only by permission of Plenum Publishing Corporation.
Homepage | More Perspectives on Science | Armchair World Directory
We appreciate any comments or questions you might have. Our e-mail address is
You can also fill out our feedback form
for sending us your comments. You may also reach us directly at
310-477-8960 in Los Angeles, California.
© Copyright 2001-2003 Armchair World