Table of Contents
Go Back
Evolution and Behavior of System Structure:
Eight Perspectives for Examining C omplex Issues
Jay Forrest
Ph.D. Student - Leeds Metropolitan University, U.K.
22375 Fossil Ridge
San Antonio, TX 78261
(210) 490-3732
systems@ jayforrest.com
Abstract
Quantitative models based on systems thinking and system science are routinely used to
explore and anticipate the likely behavior of broad and highly complex issues and
problems. While such models can provide valuable insights, they are invariably simplistic
and frequently face controversy or uncertainty in both structure and quantitative details.
The end result is that, while they may prove valuable in understanding the dynamics of
the system, their value in understanding the evolutionary and behavioral tendencies of
the system may be quite limited. A qualitative approach based upon structural
perspectives can suggest tendencies beyond the scope of quantitative models. This paper
presents eight interrelated perspectives for examining a complex issue or problem and
for inferring potential evolutionary tendencies or behavior based upon the structural
characteristics of the system under study. Experience suggests these perspectives may be
useful not only in dealing with qualitative system models, but also in validating and
troubleshooting quantitative models.
Good system dynamic models contribute to deep understanding of an issue or topic and
offer insight to those striving to understand systems or remediate problems. The literature
of system dynamics work predominantly focuses on the current situation and near-term
forecasts, where quantitative system dynamic models can be particularly accurate and
useful. As the time-horizon for understanding the system is extended, the validity of
models (system dynamics or otherwise) invariably decreases due to omitted information
and mechanisms. In order to build models that are more useful, John Sterman suggests
“..,modelers must also take care to search for and include in their models the feedback
loops, and structures that have not been important in generating dynamics to date but that
may become active as the system evolves” (Sterman 2000).
Asa long-range planning consultant, a practitioner of system dynamics, and a student in
foresight and studies of the future the author routinely addresses longer-term issues and
problems where quantitative modeling is difficult, controversial, sometimes arbitrary, and
possibly futile. The author’ s research into model failure (Forrest 2001) and work in
qualitative structural system modeling suggest that examining systems from a series of
perspectives can provide valuable insights into the evolutionary and behavioral
tendencies of systems’. These insights can suggest likely areas for shifting systems
structural features, provide logic for testing and validating both qualitative and
quantitative models, and serve as points for initiating and refining scenario planning
alternatives.
This article presents a series of individual perspectives for examining systems and system
behavior, highlights the relationship between the perspectives, and presents several brief
examples of using these perspectives to address issues. These perspectives are derived
from six fundamental concepts espoused in a number of academic disciplines:
1. Stocks drive systems (System Dynamics)
2, Feedback Loops serve as long-term (or primary) drivers of systems and provide
leverage for influencing the behavior of a system (System Dynamics, Electrical
Engineering).
3. System structure influences system behavior (System Dynamics, the work of
Michel Godet)
4. System structure patterns and evolutionary tendencies are a function of the
maturity of the system (Biology, Ecology and Biochemistry)
5. The dynamic equilibrium of an evolutionary system is a function of the stability
of the both the system and its environment (Evolutionary Ecology)
6. Fitness in a fitness landscape and the resultant pattern of possible evolution is a
function of the level of complexity of the landscape which is influenced by the
complexity of the fitness function - the number of factors determining fitness, or
in other words, the interconnectedness of fitness (Mathematical Biology and Self
Organization)
The derived perspectives provide insight into potential behavior that informs the other
perspectives, creating a cohesive framework for inferring behavioral and evolutionary
tendencies. This paper focuses on introducing the framework of perspectives and their
implications, not justifying the perspectives. Readers are referred to the references for
details regarding the perspectives and their underlying logic.
Many of the interpretations and perspectives presented in this paper will be familiar to
experienced systems thinkers but the interrelationships may not. A comprehensive
approach to the perspectives is taken to provide closure and to insure less experienced
readers do not overlook important insights that are available from applying systems
thinking. The key interpretations of these perspectives are based on observations from
system dynamics combined with quantitative research from a range of fields related to
network and system structures. As in many disciplines, the subtleties and nuances of the
relationships among the perspectives grows with ones experience and awareness.
' Evolutionary and behavioral tendencies as used in this paper refer to the shifting of the active system
structure over time and the behavioral implications of that shifting structure.
Applying these perspectives demands mental gymnastics as one zooms from macro- to
micro- perspectives, from identified causalities to externalities, from aggregated flows to
flow networks, from implicit assumptions to external uncertainties, and from the rigorous
stock-flow logic of system dynamics to intuition. Experience suggests that individuals
having substantial systems experience are likely to be more adroit at the making these
leaps. As in other qualitative approaches to systems thinking, experience with stock-flow
thinking contributes rigor to the process.
Eight Qualitative System Perspectives”
This section identifies eight perspectives for identifying system characteristics and
tendencies. The following section will show how these perspectives inform each other to
create a cohesive logic for inferring likely patterns of system behavior and evolution. The
paper closes with an example illustrating the application of these perspectives to
globalization.
System Maturity and Phases of System Evolution
While the concept of stability is frequently encountered in the literature of system
dynamics, the concept of system maturity is not. Though stability and maturity may be
related to some extent, the concept of maturity carries broader implications characterizing
the behavior of the system. System maturity is a common topic in the fields of biology,
ecology, and ecological evolution and is associated with both behavioral and evolutionary
tendencies of systems. A recommended first step in addressing a system is a simple
assessment of maturity (Troncale 2000). That initial perception is subsequently
augmented and tested as one identifies more precisely the current evolutionary phase of
the system.
A preliminary assessment of system maturity is a judgment based upon a combination of
historical and current perceptions. The answers to several simple questions guide this
assessment and provide a series of snapshots, creating a preliminary perception of
functional maturity of the system under consideration:?
e How old is the system? (Or conversely, how maturely does the system behave?)
Has the system operated through many feedback cycles?
Does the system seem to be relatively stable?
Does the structure seem to be changing? If so, how?
Is the system richly, or sparsely interconnected?
Does the number of interconnections seem to be growing or declining?
? While some of these perspectives have quantitative application they are included for their usefulness in
assessing qualitative models.
$ These questions are deliberately fuzzy to some extent. The sophistication of the answers and the
perceptions of the viewer can be expected to vary depending upon the experience and predilections of the
individual answering the questions. The cumulative response provides a basis for subsequent refinement.
Overlaps and possible conflict of the questions is addressed in subsequent text.
The author does not suggest specific criteria for determining system maturity as the
precise nature of maturity has some variation from discipline to discipline. The
judgments suggested are intuitive and benefit from experience in system analysis and
with similar systems. The goal for this preliminary perspective is to serve as a platform
for elaboration and testing with subsequent concepts and perspectives.
Mature vs. Immature Systems.
An immature system will tend to be relatively young, having survived relatively few
feedback cycles following its creation or a major disrupting event. Relationships in
immature systems should be expected to be unstable and potentially erratic. Relationships
within the system may seem somewhat chaotic and disorganized, as in many an
entrepreneurial company. One should expect new relationships to be created and tested
and existing relationships to be modified or terminated. Whereas structure for immature
systems should be expected to be transitory and evolutionary, the structure of mature
systems should be relatively stable unless other factors are impacting on the system.
The results of an assessment of maturity can be dependent on the boundaries of the
system being evaluated. A mature forest could conceivably have a constant number of
trees in every age group. From a macro perspective the forest would be stable and
mature. The constancy of numbers of older trees, however, requires the death of some
trees as they age. At a local level, the death of a tree, the creation of a hole in the canopy,
and the resulting competition to fill the void suggests a chaotic, immature system.
Stability, or instability, alone is not an adequate basis for determining the maturity of a
system. Recognizing the scalar relationships of how the system under study fits into the
overlying systems can be important in understanding the functional level of maturity.
Consistency between the state of maturity engendered by the earlier questions and the
perceived behavior of the system serves to validate the perceived level of maturity.
Inconsistency between the behavior of a system and its apparent maturity suggests other
factors are involved and invites further study.
A typical pattern of maturation involves early experimentation as new connections and
combinations are explored and tested while searching for more efficient combinations.
More mature systems should exhibit some level of efficiency and dynamic equilibrium as
the system’s age should have allowed many cycles of its various feedback loops and for
parsing of system dependencies to a relatively efficient configuration. A mature system
may experience structural change and shifting behavior, but one would directionally
expect slower, more deliberate change than in less mature systems. Systems including
longer period feedback loops should naturally be expected to take longer to display
mature stability.
An Evolutionary Pattern for Systems
Evolutionary ecologists routinely speak of a cyclical pattern of ecological evolution
consisting of four sequential phases (Ulanowicz 1997)*:
e Growth
e Development
e Maturation
e Senescence.
Economists, sociologists, and futurists frequently speak of similar patterns of evolution
though the details and character of the phases vary somewhat. Recognition of the current
phase of growth of a system provides additional insight into the likely behavior of a
system. Recognition of the evolutionary phase of a system sets expectations for not only
the current behavior of the system but also provides a basis for anticipating future shifts
in system behavior. The four evolutionary phases and their implications follow. A variety
of graphical representations are used to describe the maturation cycle focusing on
different facets of the process and using different words for the phases. Several cyles are
illustrated at the end of this topic.
The Growth Phase. The growth phase typically begins immediately after a system
undergoes a major destructive perturbation or when the system enters a new domain.
Events initiating growth phases commonly include catastrophes, technological
breakthroughs, physical relocation or expansion, and successful promotions. In the
growth phase there is little competition and necessary resources are generally adequate or
readily available. During the growth phase stocks typically display rapid or exponential
growth (within the bounds of enabling resources). The growth phase extends from
initiation to the inflection point on a typical S-curve growth pattern. During the growth
phase actors in the system generally focus more on growth and opportunity - exploiting
available resources - more than upon efficiency. The environment during the growth
phase is often turbulent, encouraging the creation and exploitation of redundant paths and
comnections to support growth and maximize utilization of available resources or
opportunities. Successful systems eventually encounter resource limitations and/or
competition for resources leading to the development phase.
The Development Phase. During the development phase growth slows - typically as a
result of declining resource availability, demand saturation, or growing competition.
Declining resources encourage efficiency and actors within the system typically begin
seeking efficiency by pruning less efficient and redundant flow paths, thereby reducing
overhead. Progression from the growth phase through the development phase is typically
slow and the transition from the growth phase to the development phase is usually more
evident after the fact than during the transition. For systems displaying the familiar S-
“ The following descriptions of the four phases generally follow the language and pattems of system
evolution described by Robert E. Ulanowicz in the book Ecology, The A scendent Perspective. Consistent
patterns of system evolution are described by other evolutionary ecologists using similar language
(Hollings 1986)(Golley 1974). Ulanowicz supports his analysis and description of the four phases with a
cohesive analysis built on a combination of thermodynamics, information theory, ecosystem energetics, and
complexity theory. Reading of Ulanowicz’s book is highly recommended to those who wish to explore the
quantitative logic underlying this paper.
shape “growth” curves, the transition from the growth phase to the development phase
would be associated with the inflection point in the growth curve. System dynamic
models show that the inflection point for simple S-curve models having rapid feedback
should be at the mid point between base level and the peak (or carrying capacity) of the
S-curve. Delays in the feedback process allow overshoot and either oscillation or collapse
depending upon the nature of the critical resource.
The Maturation Phase. In the maturation phase the system growth slows and often
peaks as resource limitations, competition, and other factors combine to restrict growth.
The path connections will have been pruned to the most efficient paths. The flow path
altematives of the system are streamlined and redundancy minimized. Overheads are
minimal as efficiency is maximized. The major system elements become tightly linked
along specific paths, leading to a fragility or brittleness of the system, making it more
susceptible to disturbances and disruptions.
Senescence. Events in senescence are highly dependent upon the nature of the system
and its environment. In absence of a major disturbance or disruption, system throughput
and activity may stabilize - i.e. the system structure will stabilize and cease to evolve.
Competition may lead to slow decline. Eventually a disruption will usually stress an eco-
system beyond the level of possible accommodation by the highly efficient flow structure
and the system will begin to decline. The speed of decline will vary with the nature of the
disruption and the characteristics and fragility of the system. While generalizations about
senescent systems are difficult, the behavior of a senescent system will range from
stagnancy to death.
Graphical Representations of Maturation Cycles. Several of the graphical
representations for the maturation cycle follow. The examples come from a variety of
sources and key on different facets of the process. While the phases may be named
differently from those above there is strong correlation to the phases described in this
paper, with Growth equating to Renewal, Development to Exploitation, Maturation to
Conservation, and Senescence to Destruction. These graphs were chosen to show how the
maturation cycle can provide insights to system characteristics other than population and
time.
Maturation Phase
Senescence
(Some possible paths
for decline)
ion
Development Phase
Growth Phase
Time
Figure 1. Development Phases for a typical S-Curve
Renewal Conservation
Ss Fon
_ 7 \ Ff \
Ey \
5 i
= Y
= “, !
AN /
=
nm ‘, /
A ™~. ral \ A
5-7 : es
Exploitation Creative
Destruction
Organization
Figure 2. Hollings suggestion for how ecosystems progress through the cycle of renewal, exploitation,
conservation, and creative destruction. The distance between successive arrows represents the
relative speed of system evolution. Adopted from Ulanowicz (1997).
Conservation
Exploitation
Renewal Destruction
Mutual Information of Flow Structure
Figure. 3 Ulanowicz's mutual information interpretation of Hollings model from Figure 2. (Used with
permission.) The process moves increasingly slowly as through the process. Destruction is typically
rapid.
Network Complexity and System Stability
Stuart Pimm (Pimm 1982) estimated that the number of effective connections? per node
in his collection of ecosystem food webs averaged about 3.1. The value of three arises
again in the work of Wagensburg, Garcia, and Sole where they suggest a “magic value of
about 3 bits per emitter [as] an actual upper limit to connectivity in real stationary
ecosystems” (Wagensburg, Garcia et al. 1990)°. The potential significance of three is
further reinforced by the work of Stuart Kauffman with Boolean networks of genetic
transitions where he found that networks remain chaotic and unstable until the
connections per node drop to about three or less, at which point the networks begin to
exhibit spontaneous, unexpected collective order (Kauffman 1991). Ulanowicz uses a
combination of Shannon diversity and information theory to calculate that the upper
boundary limit for effective connections per node in stable systems is e*”, or about 3.15
(Ulanowicz 2002).
5 Effective connections are based upon a weighting of discrete connections based on the throughput of the
highest flow. A supplier daily ordering 100 tires each from three different suppliers would have 3 effective
connections. Another ordering 200 tires from one supplier, and 80 and 20 from others would have 1.5
effective connections. Contemplating the number and magnitude of flow connections involves
disaggregating the flows in a typical system dynamics model to individual source-based flows. From a
practical perspective this would typically be a mental exercise rather than a paper or computer activity.
The term stationary is used to systems displaying maturity and stability.
These works consistently suggest that systems having more than three effective
connections per node will be unstable. Anecdotal evidence and preliminary research
confirm that a nominal limit of 3 effective connections per node may have some level of
validity in relatively mature and stable social and economic systems. In addition,
preliminary market research suggests that markets having more than three effective
suppliers will experience consolidation to approximately 3 or fewer effective suppliers. It
reaches beyond current research conclusions, but we will assume for the purposes of this
paper that immature systems having greater connectivity will be expected to organize to
value of 3.15 or fewer effective connections’ per node as they mature.
Figure 4 shows effective connectivity from 41 ecosystem studies compiled by Robert
Ulanowicz (Ulanowicz 2002). All of the studies show effective connectivity less than
3.15.
1 i i 18 eT
Topological Connectivity (m*)
Figure 4. This graph copied from Robert Ulanowicz (2001) shows the effective connectivity and
topological connectivity from 41 ecosystem studies as triangles. The dashed line separates
infeasibility (effective connections cannot exceed topological connections) from feasible. The small
dots reflect the effective connectivity of 359 randomly generated networks. The gray line at a value of
approximately 3 marks the division between stable systems (below) and unstable systems (above). All
ecosystem studies have an effective connectance of less than 3.15. (Reproduced with permission of
Robert Ulanowicz.)
Figure 5 shows the number of effective U.S. automobile manufacturers from 1896 to
1970. In the early years many companies developed vehicles but sales were dominated by
only a few. Between 1905 and 1910 the frenzied introduction of brands led the number of
7 The terms effective connections, topological connectance, network connectivity, and average mutual
information (in an information network) share common mathematical roots. Readers are referred to
Ulanowicz’s book, Ecology, The Ascendent Perspective for mathematical details.
effective manufacturers on a national basis to exceed 3°, During this period, while the
number of effective manufacturers was in the unstable range, consolidation began with
Buick, Cadillac, Oakland and Oldsmobile merging to form General Motors. By 1915
Henry Ford's inexpensive Model T acquired enough market share to bring the number of
effective manufacturers back down to the stable range. But other manufacturers began to
copy Ford’s manufacturing, and production and increased competition reduced Ford’s
market share. The number of effective manufacturers rose to 3 but consolidation pulled
the value back to the range of 2 to 2.5 where it stayed through the remainder of the
period.
Effective Automobile Manufacturers
0.5
(0)
19 ra) re) ra) 19 ra) ra)
SSSSBHRRARRIEERASRBE
DAADAAARAAAADRAGAAGD
Sage Ree eRe eae aa a as
Year
Figure 5. Profile of Effective Automobile Manufacturers in the United States.
In flow networks the strength of each flow connection is calculated relative to the largest
flow and normalized such that the greatest flow has a value of 1. The number of effective
connections is simply the sum of the normalized strengths of each flow.’ As a result if we
know the total flow and the flow from the dominant source (or the percent from the major
source) we can calculate the number of effective sources.!° The limit of 3.15 effective
sources for a stable network suggests that within a stable network the dominant flow can
have no less than 31.75 percent (1 divided by 3.15). Given that the methods in this paper
are focused on perceived system structures rather than quantified flows, this value will be
rounded off to 30 percent for the purposes of this paper.
5 During the period prior to 1915 and the introduction of the Ford Model T most auto manufacturers were
very small and sold cars locally. Many of them did not compete directly and could be argued to be
operating in different markets. As a result, the connectivity values shown are somewhat inflated and there
is some question that the limit of 3.15 flow connections was truly exceeded.
°'A node receiving flows of 40, 30, and 30 units from three sources would have 2.5 effective connections
(40/40 + 30/40 + 30/40) whereas a node receiving 80, 15, and 5 units from three sources would have 1.25
effective connections (80/80 + 15/80 + 5/80).
1° 4 node where 40 percent of the flow is from one source would have 2.5 effective connections.
Contemplating a system from the perspective of flow input diversity or connectivity
provides a basis for anticipating future shifts in network density and a logic for
anticipating which connections are likely to be pruned.
e Systems having more effective input flows than 3 (or conversely where the major
flow constitutes less than 30 percent of the total) should be expected to be
unstable.
e Systems having numerous effective input flows will evolve toward 3 or fewer.
¢ Systems will tend to prune input flows that are less efficient!’ than the
alternatives.
When visualizing the effective connectivity of a node, one should think in terms of units
of approximate equivalency. In ecological studies the flow might be carbon, or calories,
ora mineral. In economic systems the flow might be automobiles, dollars, or MM Btu of
energy equivalent. Choosing an appropriate basis can be tricky. A fox, for example, may
eat rabbits, mice, frogs, snakes, and birds. Clearly one rabbit is not equal to one frog.
There is a quality or intensity of the flow that should be taken into account. Preliminary
experience with this metric suggests that estimating the percentage of flow for the
dominant flow is a good way to estimate the flow connectivity. The level of flow
connectivity serves as a reference point for considering the environmental stability of the
system and environment as described in the following topic.
Environmental Stability and Optimal Flow Connectivity
Evolutionary ecological studies have found that environmental stability has strong
influence on the level of redundancy or overhead shown in ecosystem flow networks”,
Findings by theoretical ecologists working with flow networks reveal interesting
relationships between the level of effective connectivity in mature systems and the
stability of their external environments. These observations seem to have potential
application to systems in general.
Sole sourcing in nature is rare, except in stable, highly consistent environments, such as
rain forests where plants and animals sometimes develop single host relationships -
where a flow network would show a single connection. Flow networks in ecosystems
displaying less stability and more variation typically display more connections per node.
A number of separate studies of different mid-westem U.S. ecosystems all arrived at
average connections per node ranging from approximately 3.0 to 3.19. The optimal
connectivity in systems is clearly related to the level of turbulence."
' The perceived efficiency of a source is a function of many possible factors: quality of the flowing
material, usability of the flow material, overheads necessary to obtain or utilize that stream, and can, in
human systems at least, be purely perceptual with no physical basis.
” Flow networks trace the flow of energy, compounds, or elements through an ecosystem. Flow networks
may be more detailed, less aggregated, and more complete than typical system dynamics models.
‘3 The quantitative study of levels of turbulence - including metrics for environmental turbulence - and
optimal connectivity is one of the author's research goals for the coming year.
These insights provide a basis for modifying and informing perceptions from previous
perspectives.
e Systems that display singular, sequential inputs exhibit behavior consistent with
evolution in a relatively stable environment, display characteristics of being
relatively mature, and are likely to be relatively fragile and more susceptible to
environmental instabilities.
e Systems displaying a linear flow of singular inputs (an isolated chain) will be
viable only in relatively stable, predictable environments.
e Changes in environmental stability will shift the optimal system structure. An
increasingly stable environment will support streamlining of system structure and
increasing system efficiencies, an increasingly unstable environment will favor
increasing complexity of system structure (locate and secure new input flows and
stocks) in order to provide spare capacity and increased reliability of supply to
offset increased uncertainty of supply related to environmental instability.’
Consideration of environmental turbulence - historic, current, and anticipated - provides
a basis for modifying the behavior expected from previous characterizations and for
projecting possible structural evolutionary tendencies.
Boundary Conditions and Instability
Examination of system boundaries is broadly recognized as an important step in
validating system models. Emphasis within the field of system dynamics is typically on
insuring important feedbacks are not omitted and that exogenous variables are identified
and constraints recognized (Sterman 2000). Expanding the boundary evaluation to
include environmental stability seems prudent in view of the previous section:
¢ How stable is the external environment?
e How is the stability of the external environment changing over time?
Instability directly or indirectly related to exogenous variables serves to flag variables as
possible sources of environmental instability for the system under study. Examination of
stocks related to those variables provides additional insight into potential instabilities.
Perceptions of likely instability provide a basis for anticipating shifting connectivity
patterns within the system.
“ Increased flow connections clearly provide increased capacity and reliability only to the extent that the
new connections are independent.
Expanded Boundaries May Reveal Feedback
}-———»|
| —]
Br] 7
External
Source of i
Instability 4 |
Tight Boundaries Are Reductive
and Can Omit Feedback
Figure 6. Model boundaries influence the recognition of feedback and potential sources of
environmental instability.
Stocks and Flows
Stocks and flows are familiar to experienced system dynamicists but may need some
explanation for some readers. Stocks are the accumulators in systems. Flows are actions.
Stocks reflect what has happened historically - their levels are the product of history.
Flows reflect present conditions. If we freeze time, stocks have a level and flows cease to
exist. Recognizing stocks and flow elements in models provides insight into system
characteristics. Stocks are particularly important for they drive models (Forrester 1971)
and provide continuity through their persistence.
Rigor in stock/flow thinking is beneficial in avoiding traps inherent in interpreting causal
models. Causal loop and influence diagrams used in qualitative system dynamics are
frequently written such that the distinction of stocks and flows is not readily evident in
the elements of the diagram.
Focusing on stocks, their vulnerabilities, and stability provides insight to the stability of
the system. Flows provide the means by which stocks directly affect one another.
Recognition of stocks is important for building a stock is typically a relatively slow
process and contributes to the temporal character of a system’s dynamics. Stocks
frequently decline gradually as well, but are subject to sudden decline due to catastrophic
events. Flows are typically much more variable than stocks. The permanence of stocks
serves as a stabilizing factor in considering system dynamics and tends to create delays in
linked system behaviors.
Fluency in recognizing stocks and flows in systems and in understanding the implications
thereof is a skill that aids in analyzing system behavior. Recognition of persistence or
vulnerability of stocks provides insight into likely stabilizers and destabilizers of future
behavior.
The Concept of Enabling Stocks
Most system representations deal with only the most visible elements of the system.
Hidden assets support or enable the existence of stocks and flows in system dynamic
models and are referred to in this paper as enabling stocks. For example, in a stock-flow
depiction of a water system, a flow of water could be enabled by a riverbed, a canal, a
pipeline, and possibly pumps, depending upon the nature of the flow. A lake would be
enabled by a dam and an impermeable bed. These enabling stocks are not typically
shown in conventional systems models for they are not routinely involved in the dynamic
behavior under study. Recognition of enabling stocks provides insight into potential
vulnerabilities of the system and the reliability of these enabling stocks is important to the
continuance of historical system behavior. Failure or instability of these stocks can shift
the system into totally new behavior pattems.
A thorough review of enabling stocks requires looking at each flow and stock and asking,
“What enabling stocks support this flow (or stock)?” and in tum for each enabling stock,
“How stable and reliable is this stock and what are its vulnerabilities and limits?” This
perspective provides an enhanced sense of the environmental stability of a system and a
list of recognized candidates for wildcard events and possible turbulence.
In practice, the examination of every variable in a causal diagram for implied stocks and
flows and those, in retum, for implied enabling stocks, is likely to require more effort
than is practical. A spot check of enabling stocks for those variables perceived as most
important, vulnerable, or where instability is of concern seems prudent when examining
the potential boundaries of the model. Sensitivity to the importance of enabling stocks
also serves as a potential flag for problem areas during the development of both
qualitative and quantitative system models.
Delays
Delays are powerful features in real-world systems and networks that are frequently
minimally considered in qualitative systems models. Often all causalities are treated
equally and shown as an arrow. Some practitioners indicate longer delays by putting a
break in an arrow. Very long delays relative to the time frame of the model are frequently
omitted either through lack of recognition or because they are deemed not pertinent to the
time horizon of the model. Contemplation of delays and their impacts offers additional
insights into possible behavior of the system under study. Recognition of delays -
whether inherent, omitted, explicit, or hidden - and their potential magnitude and period
provide a basis for considering the potential dynamics (and maturity) of the system.
Discussing delays in detail should be familiar to practitioners of system dynamics and is
beyond the scope of this paper. Readers unfamiliar with the following concepts are
referred to the book Business Dynamics (Sterman 2000) for an excellent overview of
material and information delays and a discussion of variable vs. constant delay time
periods.
A delay occurs when the output of a process lags the input. Delays permeate all systems
via accumulations, processing, and information delays. Delay length often plays a role in
deciding what causalities and relationships to include while setting system boundaries for
amodel.
All delays involve at least one stock with the nature of the delay depending upon the
characteristics of the stock. Conveyors, ovens, and accumulating stocks all contribute
different dynamics to downstream processes. Information delays result from the fact that
it takes time to receive and process data and to subsequently act on that data. If
information delays are relatively long, they may complicate recognition of system
characteristics and causalities. Information delays can also interfere with system control
and response, contributing to system instability. Variable delays will also influence the
dynamics of a system and potentially aggravate instabilities.
Once again, experience with system dynamic models proves valuable to the qualitative
modeler. Recognition of the delays in a model, the relative magnitudes of the delays, the
characteristics of the delays, and the implicit boundaries of the model with respect to
delays provide insight into the levels of variability within the model and provides a basis
for contemplating the boundaries of the model, for evaluating information about the
systems behavior, and for anticipating potential response and control issues.
Feedback Structures”
The concept of feedback structures is fundamental to system dynamics. Disturbances
ripple through nonrecursive linear systems with no enduring influence on behavior.
Feedback loops provide mechanisms for current conditions to influence future conditions
and frequently to dominate long-term system behavior. Two important characteristics of
feedback loops are their type - positive or negative - and delay period.
Short period feedback loops tend to be visible and dominate short-term behavior. Slower,
longer-term feedback loops are frequently less visible as human perception of causality
encounters difficulty in connecting cause and effect relationships that are separated in
time and space (Senge 1990). Careful attention must be given to identifying longer-term
feedback structures when addressing moderate- and longer-term topics and issues such as
unintended consequences. Recognition of feedback structures suggests avenues for
potential future behavioral departure from current trends. Recognition of the delay period
*S Readers unfamiliar with feedback structures and feedback polarity are specifically referred to John
Sterman’s discussion of feedback and feedback polarity in Business Dynamics.
of the feedback structure provides a basis for anticipating the temporal impacts of the
feedback loop.
Characterization and recognition of a feedback loop as positive or negative is beneficial
when possible, but is often difficult in highly complex systems where a long sequence of
fuzzy and possibly interrelated webs of feedback are involved. Sterman states that
ambiguity when identifying the polarity of causal links suggests that multiple conflicting
causal paths are hidden within the arrow of causality and that they should be broken out
until unambiguous polarity can be assigned to the causal arrow (Sterman 2000). This
approach is helpful in clarifying causal loop structures and mechanisms. For causal
diagrams where the polarities can be established one should expect to find several to
many negative feedback loops for every positive feedback loop. Failure to offset positive
feedback loops with negative feedback loops is a clear sign that the model is reductive
and that potential mechanisms for future behavioral shifts have been excluded.
Brainstorming of potential negative feedback loops provides a basis for identifying
overlooked and potential mechanisms and thereby suggesting potential unintended
consequences.
Focusing on feedback structures, their period, and their polarity provides insight to the
overall dynamics of the system, a sense of the temporal dynamics of the system, a basis
for recognizing missing and potential elements and structures, and for considering
unintended consequences.
Structural Dependency
When evaluating a system for robustness and for potential weaknesses, robustness will be
maximized when flow sources (both stocks and enabling stocks) are fully independent. In
practice, full independence is rare. Most sources share some attributes and dependencies
that serve as potential vulnerabilities to the system. Items such as shared infrastructure
(roads and electric power grids for example), common upstream sources, and
technological standards create commonalities across sources that create multi-source
vulnerability to singular events or shifts.
This perspective augments the perspective created by stocks, flows, and enabling stocks
by encouraging recognition of the structural dependencies - the shared vulnerabilities -
of the system. The perspective of structural dependency is based on a reductionist view of
the stocks and flows, by burrowing down into their dependencies and recognizing shared
dependencies across sources.
Fitness Complexity
The concept of macro-interdependency as used in this paper is related to the overall
complexity of the dependencies of a node.'® Within this perspective the fitness of a node
in a system will be a function of the number of dependencies of that actor. The fitness of
a truly independent node would be fixed. The work of Stuart Kauffman with fitness
landscapes shows that the nature of the fitness landscape - a map of fitness as a function
of the condition of the dependent variables - and of the evolutionary alternatives
available shift as the number of variables is increased. Nodes with relatively simple
fitness formulas have large areas of high fitness (or viable possibilities) and the map is
smooth. Evolution from low fitness to higher fitness is possible, as the smoothness of the
fitness map allows incremental improvement toward the peak fitness. As the complexity
of the fitness formula increases (i.e. the number of variables increase) the fitness map
tends to be squeezed to a plane - the areas of high fitness shrink as more factors must be
“good” for overall fitness to be high. And, perhaps more importantly, the fitness
landscape becomes increasingly rough - with low peaks separated by cliffs and valleys.
The roughness of complex fitness landscapes generally denies the ability to evolve
incrementally toward high fitness as small increments of change generally only moves
the actor toward a low, nearby peak or - if on the peak - off of a peak to a lower fitness.
In such a landscape major mutations (major changes, new ideas) are the only viable
method for escaping a local peak.
This perspective suggests three key insights for testing and informing other perspectives:
e Actors in systems displaying increasing interdependence can be expected to find
it increasingly difficult to maintain fitness widely different from the average (as
the fitness map squeezes toward a plane).
e Actors in systems displaying increasing interdependence are likely to become
increasingly fragmented (due to the increasingly rough fitness landscape),
e Incremental progress or movement in an increasingly interdependent system
should be expected to grow increasingly difficult and the need for drastic changes
will grow (also related to the roughness of the fitness landscape)
Linking Qualitative System Perspectives
The perspectives described mutually inform each other to create a cohesive approach for
anticipating behavioral and evolutionary tendencies of a system under study. This section
presents an overview of the primary relationships among these perspectives. Space
restrictions allow consideration of only the most obvious relationships. Experience and
‘© Tn contrast to flow connectivity, which focuses on singular items flowing to a node, the fitness
complexity reflects the number of items flowing to (or needed by) the node. A node having greater
numbers of needs will have a more complex fitness function and thereby a higher fitness complexity.
familiarity with the perspectives and their insights is likely to suggest more subtle
relationships between the perspectives.
Fitness
Complexity
Network [ ~
Connectivity ~
N
\
System Maturity
ry \
NG
Environmental \
!
1 im - ~ _
i wy] Stability be ns !
Z + oe ‘ Structural
1 / . Dependency
17 I \ ry
1 ! \ /
' I 1 v /
» F Y Stocks and
ystem Flow
Boundaries ' “
k a
| /
\ \ 7)
XN
x Y a “
> py} Feedback jae
Structures
Figure 7. Proposed logic for linking system perspectives
The core logic of the proposed perspective approach to systems behavioral and
evolutionary tendencies lies in the core relationships between perceived system maturity,
environmental stability, and network connectivity. Observation provides insights and
expectations with respect to the state of maturity of the system under study. Examination
of the relationships within the system provides insight to the existing state of effective
connectivity in the system. Perception of inconsistencies between the perceived maturity,
existing connectivity, and current and historic environmental instability suggest further
exploration and examination of the model are appropriate.
Consideration of the other perspectives informs the model definition and evaluation. For
example, the model boundaries and feedback structures provide insight to the level of
maturity and the possible nature of cyclic feedback phenomena on environmental
stability. Examination of stocks and flows further contributes to the perception of
environmental stability and provides insight to the structural dependency and the
robustness of the system components. These in turn provide additional insight to the
environmental stability and network connectivity of the system. Persistent inconsistencies
in the perspectives may indicate that the system is in transition as changes in
environmental factors are creating turbulence and structural stress and/or change.
Comparing environmental turbulence and network connectivity over the historical time
has proven useful in resolving persistent inconsistencies.
Once the historic and current perspectives are understood, focus can shift forward to the
future. Recognition of trends and vulnerabilities related to stocks provides a basis for
elaborating on the feedback structures and boundaries of the model, of increasing or
decreasing environmental turbulence, and of trends in connectivity.
A Brief Example
Globalization is frequently referred to in terms of increasing communication and trade
around the world. The long-term impact of globalization remains uncertain with
interpretations ranging from homogenization to mega-corporations. Applying the
perspectives suggests alternative impacts and directions for global evolution.
For the purposes of this example, the process begins by contemplating the state of the
planet from the perspective of the United States with respect to system maturity. Network
flow connectivity and environmental turbulence are then considered. The boundaries
throughout this example are the planet earth. Historic trends are considered. The current
state will be assessed. And, the impact of increasing global trade and interdependence
and increasing communications will be considered. Differing experiences and perceptions
could easily lead readers to a different descriptions and conclusions. Some of the
interpretations that follow may stimulate some controversy. Surfacing these differences
in-group processes provides avenues for closing perceptual gaps among group members.
In this example, the interpretations are used to illustrate not only how the perspectives
can work and mutually inform each other, but also to use selected perceptions to support
the validity of the perspectives as a useful tool. The discussion is deliberately general - at
an overview level. Readers are encouraged to fill in the gaps and develop their own
understanding of the logic presented. Within the presented framework, assessments of
specific industries and regions where factors such as the number of effective suppliers
can be estimated can be much more specific than the general overview that follows.
The Current Situation
Global population is soaring. While it is debatable whether or not we have exceeded
sustainable populations and economic activity, signs of human induced environmental
stress and possible resource shortages are clearly visible. Though the booming population
might indicate the planet is still in the growth phase from a human perspective, the clear
stresses and looming shortages suggest the development phase, possibly approaching the
point of restricted growth that characterizes maturation. The growth of human population
over history is arguably dominantly attributable to improvements in technology related to
food production and health care. Improvements in technology have led to incremental
increases in longevity and infant survival that have driven the growth in population.
These improvements which act as environmental turbulence have come at a rate that has
precluded equilibrium - i.e. new increases in longevity and infant survival have arrived at
a rate that is leading population to grow at an exponential rate. At the same time,
technology has mitigated the negative potential impacts of resource limitations such as
food and resources. Thus the overall planetary state might be described as developmental,
with technology encouraging growth, and resource issues threatening to plunge the planet
into maturity, i.e. the disruptions and opportunities of technologic improvement have
offset or delayed the limitations of resources, allowing continued growth. The population
is in a state of dynamic quasi-equilibrium between these two factors. Continued growth
depends upon continued technological progress with the familiar “overshoot and
collapse” alternative should technology fail to overcome the looming shortages.
Historically the global economy has been a patchwork of local economies, beginning
with families and tribes and evolving through fiefdoms and nations to an increasingly
interconnected web of international trade and dependency. Technologies of
communication and transportation have combined with political policies to enable and
encourage international trade, particularly following the proliferation of computers and
the Internet in the 1990s. Over the past ten years global communication and
transportation has become practical at virtually all levels, enabling every possible
international connection. The creation and exploration of the interests and synergies of
these connections has been a major contributor to turbulence in the economic and
business sectors as manufacturing, jobs, and trade issues shape and reshape business
relationships. Experimentation reigns as new connections - countries, suppliers, and
customers - are tested and potentially cast aside as new connections are tested. From a
“final form” perspective the global economy is clearly still in its early infancy and growth
phase.
The youthful, exploitative, behavior of the global environment provides an interesting
contrast to the state of US businesses. Markets and opportunities in the United States are
more mature, and arguably, on the whole, in the developmental or maturation phases.
Competition is relatively intense and pressures for profits are high. The govemment and
business have combined to create a stable, predictable environment for business in the
US. Reliable power and distribution systems combined with technological advances to
enable a strong push for optimizing flow paths surrounding a company, focusing on the
most efficient and profitable alternatives. Concepts such as ISO 9000, single sourcing,
and just-in-time supply strategies recently emerged to minimize overhead and maximize
profits. The perspective of environmental stability and optimal flow connectivity suggests
that such strategies are only viable in extremely stable environments.
From a social perspective we have seen communication ability grow progressively
through the past two hundred years with the pony express, the telegraph, telephone, radio
and television. Over recent years communication has flourished with the cell phones,
email, and the Internet. We communicate not only with those near us - family, friends
and neighbors - but also with strangers whom we have never met, via list serves,
websites, and email. While some have projected global communication will create
homogenization, the perspective from evolution in a fitness landscape suggests increasing
interconnectivity would lead to increasing fragmentation and a leveling of the fitness
landscape. Such seems to be the case. Ideas, concepts, rumors, lies, and facts inundate us.
The rising flow of information encourages a portion of society to listen selectively and to
reject foreign and unfamiliar concepts. Communication within groups is reinforcing and
supporting group thinking, leading to single issue political and action groups, and
encouraging isolated extremist thinking and even terrorism. The trend toward single
focus groups seems to support Kauffman’ s suggestion that increasingly complex fitness
factors promote fragmentation. Kauffman’ s suggestion that fitness levels are squeezed
toward a plane as fitness complexity rises is less clearly visible in historic trends. While
wealth and educational gaps seem to be rising, virtually all US citizens have telephones,
radios, televisions, and other amenities and have similar levels of access to information.
An increasing level of interdependence has been developing across the scale of
globalization. The health of national economies is increasingly dependent upon the health
of other economies (such as the United States dependence upon Mexico and vice versa).
Within countries, the sharing and cross connecting of utilities make not just the local
level vulnerable to a blackout in the event of a local problem but expand that
vulnerability across regions and perhaps the entire nation.
The combined perspectives highlight a number of tensions and their structural roots.
Among those are:
e The tension of technological advancement and resource depletion
e The tension of a relatively mature planet from a resource perspective
accommodating the turbulence of immature globalization and connectivity
e The tension of mature industries pursuing highly efficient sourcing strategies in
an increasingly interdependent and therefore turbulent global environment
e The tension of fragmentation and confrontation of ideas and movements
Social, political and business entities and people are struggling to deal with the shifting
possibilities, opportunities, challenges, and threats as new connections are enabled,
recognized, activated, and pruned, and complain of the “challenge of keeping up”, the
“volume of information and communication”, and of “the pace of change” suggesting
that the systems and methods they are using are not accommodating the current level of
environmental turbulence.
Key Assumptions Regarding the Future
The perspectives fit well into scenario planning, contributing logic for interpreting
altemative assumptions. To maintain brevity this example will consider only one case.
For this example it is assumed that globalization continues with international
comnections, business activities, communication, and interdependency growing.
Impacts of Continuing Globalization
The perspective of fitness complexity suggests that the implications of increasing
interconnectivity are fragmentation, increasing environmental turbulence, and a
compression in the range of possible fitness. These implications are examined briefly and
the impacts traced through other perspectives.
Globalization is still in its relative infancy and growth phase as new connections are
created and tested. Substantial turmoil remains as countries, companies, political
movements, and even individuals struggle to deal with their new relationships,
opportunities, threats, and abilities. Longer term, turmoil should subside as relationships
stabilize.
Increasing fragmentation is to be expected over the near term - in the form of single topic
political blocks, lifestyle alternatives, and other social/political phenomena - as the
breadth of human personality, interests, and experience lead to differing perspectives and
goals. The number of manufacturers of globally common products may defy the
fragmentation phenomenon as global companies penetrate international markets, grow
more ubiquitous, and local altematives pare down to 3 effective suppliers that will be
global rather than local in profile. The nature of products remains elusive. Pursuit of
efficiency should lead to global homogenization and commoditization of mass products
but technology is enabling mass customization of products such as custom fit blue jeans
and bicycles. Human nature seems to favor customization and uniquity suggesting that
over the long-term mass customization might be the favored path.
Increasing turbulence can be expected to increase the overhead of society, business, and
government as disruptions ripple through local and global economies. Events such as the
recent New England blackout, acts of terrorism, and simple human errors threaten to
impact our lives. Technology, security, and procedures will have increased focus on
reducing the frequency of disruptive events and on minimizing the ripple effect.
Turbulence will stimulate movement toward multiple independent suppliers - an increase
in the average effective connectivity of our supply infrastructures implying a lowering of
efficiency as a result of the overhead penalties of redundancy and independence. The
logic of supply chains in the United States and the focus on dedicated, just-in-time supply
are likely to encounter difficulties in a turbulent world. One would expect the typical
number of effective suppliers in specific markets to rise toward the limit of 3.2 in many
business areas and for critical necessities.
In a world of increasing globalization one would anticipate that global companies will
increasingly seek access to local markets. Local suppliers should be expected to struggle
in commodity businesses as larger, more efficient, and deeper pocketed competitors offer
more for less. The possibility of mass customization could further shift the potential
advantage to global suppliers. Two potentially countertrends offer potential insight into
the ultimate balance between global and local supply:
1. Rising energy prices which would influence the competitiveness of shipping
goods, thus making local supply more attractive - particularly for bulkier, heavier,
and more fragile products
2. Virtualization, nanotechnology, and technology are combining in many areas to
reduce the costs of local production. This trend has mixed benefits, however, as
virtualization and nanotechnology are also making many products more easily
deliverable.
In any event, it seems likely that in many markets international companies will continue
to expand and acquire or bankrupt local suppliers.
Growing intercomnectivity and interdependence imply that reductionist perspectives and
problem solving approaches will grow less effective at solving the problems facing
government, businesses, and individuals. Holistic approaches and systems thinking
should gain favor. Techniques for aiding humans in recognizing and dealing with
multiple issues and perspectives, feedback issues, and unintended consequences should
be increasingly valued. Education will need to shift from simplistic answers to logic and
to handling complexity. Design of organizations and infrastructures should balance
efficiency with independence and robustness. This implies a massive shift in educational
focus and philosophy for the bulk of the U.S. educational system.
Conclusion
This paper presents logic for merging insights from a variety of systems paradigms into a
cohesive approach for inferring behavioral and evolutionary tendencies of systems based
predominantly upon qualitative assessments of system structure. The logic presented is
but one approach to qualitative system studies and resides in a matrix of facilitation,
problem solving, and other system-related logics and perspectives. Extension and
refinement of the logic is desirable and practical. A pplication of Michel Godet’s
MICMAC methodology (Godet 1999), for example, offers the benefit of suggesting
leverage points for shifting system behavior, without requiring quantification of complex
issue models. Incorporation of insights from other systems oriented disciplines offers the
potential of broader applicability and stronger inferences. Integration of this logic into
Geoff Coyle’s more comprehensive ACTIFELD approach (Coyle 2004) seems to offer
potential benefits. The ultimate goal for this paper is that it stimulates new thinking and
ideas in other systems thinkers that will lead to stronger methodologies and solutions.
Bibliography
Coyle, G. (2004). Practical Strategy: Structured Tools and Techniques. Essex, England,
Pearson Educated Limited.
Forrest, C. J. (2001). Time Related Sources of Model Failure. System Dynamics Society
International Conference.
Forrester, J. W. (1971). Principles of Systems. Cambridge, Massachusetts USA,
Productivity Press.
Godet, M. (1999). Scenarios and Strategies, A Toolbox for Problem Solving. Paris,
Laboratory for Investigation in Prospective Strategy and Organization. 2001.
Golley, F. B. (1974). Structural and functional properties as they influence ecosystem
stability. Intemational Congress of Ecology, Wageningen, Netherlands, Centre for
Agricultural Publishing and Documentation.
Holling, C. S. (1986). The resiliance of terrestrial ecosystems: Local surprise and global
change. Sustainable Development of the Biosphere. W. C. Clark and R. E. Munn.
Cambridge, Cambridge University Press.
Kauffman, S. A. (1991). “Antichaos and adaption.” Scientific A merican(265): 78-84.
Pimm, S. L. (1982). Food Webs. London, Chapman and Hall.
Senge, P. M. (1990). The Fifth Discipline: The Art and Practice of the Learning
Organization. New Y ork, Doubleday/Currency.
Sterman, J. D. (2000). Business Dynamics: Systems Thinking and Modeling fora
Complex World. Boston, McGraw-Hill.
Troncale, L. R. (2000). Personal communications.
Ulanowicz, R. E. (1997). Ecology, The A scendent Perspective. New Y ork, Colombia
University Press.
Ulanowicz, R. E. (2002). “The balance between adaptability and adaptation.”
BioSystems(64): 13-22.
Wagensburg, J., A. Garcia, et al. (1990). “Connectivity and information transfer in flow
networks: Two magic numbers in ecology?".” Bull. of Math. Biol.(52): 733-740.
Back to the Top
CC BY-NC-SA 4.0