Write a new introduction and a new conclusion!!


I did a summury for an essay, but the professore asked me to write a new introduction and a new conclusion

for the summury I did. So if u think u can help with that please contact me. You have to read the whole essay first so you have better understanding about it and then read the summery I did, after that you only need to write a NEW INTRODUCTION AND CONCLUSION IN A BETTER WAY.






Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.


INFORMS is collaborating with JSTOR to digitize, preserve and extend access to Organization Science.


This content downloaded from on Wed, 4 Sep 2013 16:13:49 PM
All use subject to JSTOR Terms and Conditions

Complexity Theory and Organization Science

Philip Anderson
Amos Tuck School, Dartmouth College, Hanover New Hampshire 03755-9000

Complex organizations exhibit surprising, nonlinear behavior.
Although organization scientists have studied complex organi-
zations for many years, a developing set of conceptual and com-
putational tools makes possible new approaches to modeling
nonlinear interactions within and between organizations. Com-
plex adaptive system models represent a genuinely new way of
simplifying the complex. They are characterized by four key
elements: agents with schemata, self-organizing networks sus-
tained by importing energy, coevolution to the edge of chaos,
and system evolution based on recombination. New types of
models that incorporate these elements will push organization
science forward by merging empirical observation with com-
putational agent-based simulation. Applying complex adaptive
systems models to strategic management leads to an emphasis
on building systems that can rapidly evolve effective adaptive
solutions. Strategic direction of complex organizations consists
of establishing and modifying environments within which ef-
fective, improvised, self-organized solutions can evolve. Man-
agers influence strategic behavior by altering the fitness land-
scape for local agents and reconfiguring the organizational
architecture within which agents adapt.
(Complexity Theory; Organizational Evolution; Strategic

Since the open-systems view of organizations began to
diffuse in the 1960s, comnplexity has been a central con-
struct in the vocabulary of organization scientists. Open
systems are open because they exchange resources with

the environment, and they are systems because they con-
sist of interconnected components that work together. In
his classic discussion of hierarchy in 1962, Simon defined
a complex system as one made up of a large number of
parts that have many interactions (Simon 1996).
Thompson (1967, p. 6) described a complex organization
as a set of interdependent parts, which together make up
a whole that is interdependent with some larger environ-

Organization theory has treated complexity as a struc-
tural variable that characterizes both organizations and
their environments. With respect to organizations, Daft
(1992, p. 15) equates complexity with the number of ac-
tivities or subsystems within the organization, noting that
it can be measured along three dimensions. Vertical com-
plexity is the number of levels in an organizational hier-
archy, horizontal complexity is the number of job titles
or departments across the organization, and spatial com-
plexity is the number of geographical locations. With re-
spect to environments, complexity is equated with the
number of different items or elements that must be dealt
with simultaneously by the organization (Scott 1992, p.
230). Organization design tries to match the complexity
of an organization’s structure with the complexity of its
environment and technology (Galbraith 1982).

The very first article ever published in Organization
Science suggested that it is inappropriate for organization
studies to settle prematurely into a normal science mind-
set, because organizations are enormously complex (Daft
and Lewin 1990). What Daft and Lewin meant is that the
behavior of complex systems is surprising and is hard to

ORGANIZATION SCIENCE/Vol. 10, No. 3, May-June 1999 Copyright ? 1999, Institute for Operations Research
pp. 216-232 and the Management Sciences

This content downloaded from on Wed, 4 Sep 2013 16:13:49 PM
All use subject to JSTOR Terms and Conditions

PHILIP ANDERSON Complexity Theory and Organization Science

predict, because it is nonlinear (Casti 1994). In nonlinear
systems, intervening to change one or two parameters a
small amount can drastically change the behavior of the
whole system, and the whole can be very different from
the sum of the parts. Complex systems change inputs to
outputs in a nonlinear way because their components in-
teract with one another via a web of feedback loops.

Gell-Mann (1994a) defines complexity as the length of
the schema needed to describe and predict the properties
of an incoming data stream by identifying its regularities.
Nonlinear systems can difficult to compress into a par-
simonious description: this is what makes them complex
(Casti 1994). According to Simon (1996, p. 1), the central
task of a natural science is to show that complexity, cor-
rectly viewed, is only a mask for simplicity. Both social
scientists and people in organizations reduce a complex
description of a system to a simpler one by abstracting
out what is unnecessary or minor. To build a model is to
encode a natural system into a formal system, compress-
ing a longer description into a shorter one that is easier
to grasp. Modeling the nonlinear outcomes of many in-
teracting components has been so difficult that both social
and natural scientists have tended to select more analyt-
ically tractable problems (Casti 1994). Simple boxes-and-
arrows causal models are inadequate for modeling sys-
tems with complex interconnections and feedback loops,
even when nonlinear relations between dependent and in-
dependent variables are introduced by means of expo-
nents, logarithms, or interaction terms. How else might
we compress complex behavior so we can comprehend

For Perrow (1967), the more complex an organization
is, the less knowable it is and the more deeply ambiguous
is its operation. Modem complexity theory suggests that
some systems with many interactions among highly dif-
ferentiated parts can produce surprisingly simple, pre-
dictable behavior, while others generate behavior that is
impossible to forecast, though they feature simple laws
and few actors. As Cohen and Stewart (1994) point out,
normal science shows how complex effects can be un-
derstood from simple laws; chaos theory demonstrates
that simple laws can have complicated, unpredictable
consequences; and complexity theory describes how
complex causes can produce simple effects.

Since the mid-1980s, new approaches to modeling
complex systems have been emerging from an interdis-
ciplinary invisible college, anchored on the Santa Fe In-
stitute (see Waldrop 1992 for a historical perspective).
The agenda of these scholars includes identifying deep
principles underlying a wide variety of complex systems,
be they physical, biological, or social (Fontana and

Ballati 1999). Despite somewhat frequent declarations
that a new paradigm has emerged, it is still premature to
declare that a science of complexity, or even a unified
theory of complex systems, exists (Horgan 1995).
Holland and Miller (1991) have likened the present sit-
uation to that of evolutionary theory before Fisher devel-
oped a mathematical theory of genetic selection.

This essay is not a review of the emerging body of
research in complex systems, because that has been ably
reviewed many times, in ways accessible to both scholars
and managers. Table 1 describes a number of recent,
prominent books and articles that inform this literature;
Heylighen (1997) provides an excellent introductory bib-
liography, with a more comprehensive version available
on the Internet at http://pespmcl.vub.ac.be/
Evocobib. html. Organization science has passed the
point where we can regard as novel a summary of these
ideas or an assertion that an empirical phenomenon is
consistent with them (see Browning et al. 1995 for a path-
breaking example).

Six important insights, explained at length in the works
cited in Table 1, should be regarded as well-established
scientifically. First, many dynamical systems (whose state
at time t determines their state at time t + 1) do not reach
either a fixed-point or a cyclical equilibrium (see Dooley
and Van de Ven’s paper in this issue). Second, processes
that appear to be random may be chaotic, revolving
around identifiable types of attractors in a deterministic
way that seldom if ever return to the same state. An at-
tractor is a limited area in a system’s state space that it
never departs. Chaotic systems revolve around “strange
attractors,” fractal objects that constrain the system to a
small area of its state space, which it explores in a never-
ending series that does not repeat in a finite amount of
time. Tests exist that can establish whether a given pro-
cess is random or chaotic (Koput 1997, Ott 1993). Sim-
ilarly, time series that appear to be random walks may
actually be fractals with self-reinforcing trends (Bar-Yam
1997). Third, the behavior of complex processes can be
quite sensitive to small differences in initial conditions,
so that two entities with very similar initial states can
follow radically divergent paths over time. Consequently,
historical accidents may “tip” outcomes strongly in a par-
ticular direction (Arthur 1989). Fourth, complex systems
resist simple reductionist analyses, because interconnec-
tions and feedback loops preclude holding some subsys-
tems constant in order to study others in isolation. Be-
cause descriptions at multiple scales are necessary to
identify how emergent properties are produced (Bar-Yam
1997), reductionism and holism are complementary
strategies in analyzing such systems (Fontana and Ballati

ORGANIZATION SCIENCE/Vol. 10, No. 3, May-June 1999 217

This content downloaded from on Wed, 4 Sep 2013 16:13:49 PM
All use subject to JSTOR Terms and Conditions

PHILIP ANDERSON Complexity Theory and Organization Science

Table 1 Selected Resources that Provide an Overview of Complexity Theory

Allison and Kelly, 1999 Written for managers, this book provides an overview of major themes in complexity theory and
discusses practical applications rooted in-experiences at firms such as Citicorp.

Bar-Yam, 1997 A very comprehensive introduction for mathematically sophisticated readers, the book discusses the
major computational techniques used to analyze complex systems, including spin-glass models,
cellular automata, simulation methodologies, and fractal analysis. Models are developed to describe
neural networks, protein folding, developmental biology, and the evolution of human civilization.

Brown and Eisenhardt, 1998 Although this book is not an introduction to complexity theory, a series of small tables throughout the
text introduces and explains most of the important concepts. The purpose of the book is to view
strategic change through the lens of complexity theory. Brown and Eisenhardt contend that
successful companies manage in turbulent environments by charting a strategic course that puts
them on the edge of chaos, poised between order and disorder.

CalResCo Introduction to Available at http: / /www. calresco. org/intro. htm#def, this website provides a comprehensive
Complex Systems introduction and set of links that introduce many different aspects of complexity theory. It is a good,

free, and relatively nontechnical starting place for exploring the topic.
Capra, 1996 Written for laymen, this book traces the development of systems theories, then discusses self-

organization and the mathematics of chaos theory. Complexity theory is discussed in the context of
how life emerged from inert chemical components.

Coveney and Highfield, 1995 Written for laymen, this book traces the history of thinking about computational complexity in
mathematics. It discusses quite readably cellular automata, spin-glass models, neural networks,
and genetic algorithms, self-organization, artificial life, and theories of brain functioning.

Cowan, Pines, and Meltzer, 1994 This is a collection of papers presented at the Santa Fe Institute’s Fall 1991 workshop on integrative
themes of the sciences of complexity. Although no single chapter provides a comprehensive
overview, taken together, the chapters thoroughly cover the predominant themes developed by
scholars of complex systems. Six chapters introduce fundamental concepts; the rest provide a
number of examples of complex adaptive systems (principally, but not exclusively, biological), while
four explore cellular automata, self-organized criticality and the “edge of chaos,” and the concept of

Mainzer, 1994 Written at a high level but without extensive use of mathematics, this is a comprehensive overview of
complex systems theory. Chapters describe different models in the context of the evolution of
matter, the evolution of life, the evolution of the brain, the evolution of artificially intelligent
computational systems, and the evolution of human society.

United Nations University, 1985 A collection of chapters from a very early conference on complexity theory in 1984, this book no
longer captures the main lines along which complexity theory has developed. Nonetheless, the
individual chapters, though quite eclectic, remain thought-provoking. This book is more a source of
interesting ideas than a comprehensive introduction to the field.

Waldrop, 1992 Written for laymen, this book is a popular yet sophisticated introduction to complexity theory. It is the
most readable introduction to the field for nonspecialists. In tracing the founding and early years of
the Santa Fe Institute, Waldrop touches on applications of complexity theory to economic systems,
Boolean networks and the NK model, genetic algorithms and classifier systems, self-organization
and artificial life, and evolution to the edge of chaos.

Weisbuch, 1991 This is a methodological book, which introduces analytical techniques at the introductory graduate-
school level. Topics include cellular automata, neural networks, simulated annealing, Boolean
networks, and evolutionary population dynamics.

1999). Fifth, complex patterns can arise from the inter-
action of agents that follow relatively simple rules. These
patterns are “emergent” in the sense that new properties
appear at each level in a hierarchy (Holland 1995). Sixth,
complex systems tend to exhibit “self-organizing” behav-
ior: starting in a random state, they usually evolve toward
order instead of disorder (Kauffman 1993).

The Evolution of Modern Complexity
As Simon (1996) has pointed out, these ideas have deep
historical roots; the intellectual ferment reviewed by the
works in Table 1 represents the third wave of interest in
complex systems this century. First, the years after World
War I saw an explosion of interest in holism and gestalt

218 ORGANIZATION SCIENCE/VOl. 10, No. 3, May-June 1999

This content downloaded from on Wed, 4 Sep 2013 16:13:49 PM
All use subject to JSTOR Terms and Conditions

PHILIP ANDERSON Complexity Theory and Organization Science

theories. Then, cybernetics and general systems theory
emerged after World War II, fueled by the success of
wartime feedback-control devices and accelerated by the
development of computers. These intellectual movements
meant to replace reductionism with an appreciation for
modeling interactions instead of simplifying them away.
Presaging today’ s most enthusiastic “science of complex-
ity” boosters, one of the founders of cybernetics declared
in 1955: “Science is at last giving serious attention to
systems that are intrinsically complex” (Ashby 1981, p.

Cybernetics emphasized coordination, regulation, and
control using feedback loops (Ashby 1956). General sys-
tems theory (Forrester 1961, von Bertalanffy 1968) at-
tempted to elucidate deep principles underlying all types
of systems whose components are linked by feedback
loops. Both influenced the intellectual revolution that
swept organization theory in the 1960s and ushered in a
view of organizations as open systems (Katz and Kahn
1978). The systems design school of organization theory
(Haberstroh 1965) was based on a characterization of sys-
tems as a collection of black boxes connected by input-
output loops. A new breed of systems analysts designed
work processes and organizational control systems
around the ideas of systems theory and cybernetics
(Beniger 1986). Pointing out that social organizations are
more loosely coupled than most physical systems, Weick
(1979) introduced a theory of organizing based on loosely
coupled subassemblies called “double interacts,” behav-
ioral cycles linking the behavior of two people in a set of
feedback loops. Systems dynamics models continue to
inform a broad stream of contemporary research (e.g.,
Samuel and Jacobsen 1997, and Sterman and Witterberg
in this issue), and nonlinear dynamical systems theory has
been used to study many complex, nonlinear behavior
patterns (Epstein 1997).

According to Simon (1996), the third wave of theories
about complex systems is rooted in a new understanding
of equilibrium that emerged in the late 1960s. Catastrophe
theory (Thom 1975) explained how in some deterministic
systems, a small shift in a parameter could send the sys-
tem to a very different equilibrium. Chaos theory ex-
plored how some dynamical systems that appear to be
random are, in fact deterministic (Thietart and Forgues
1995). Typically, such systems take the value of a vari-
able in time t, stretch it, then fold it to produce a new
value at time t + 1 (Cohen and Stewart 1994). The
stretching operation magnifies small initial differences,
while the folding operation constrains the range of values
to a relatively small volume of the state space. It is usually
impossible to forecast the exact value of a chaotic system
in nature, because small measurement errors between two

‘apparently identical values at time t can lead to large
differences at time t + 1. However, such systems are in
equilibrium around a strange attractor, a limited region of
the state space within which the system stays perma-
nently. Similarly, nonchaotic dynamical systems that ap-
pear to be random (“white noise”) may in fact have un-
derlying structural time trends (“colored noise”) as
Dooley and Van de Ven’s paper (this issue) discusses.
Colored noise can have important effects on outcomes,
such as the risk that a population will become extinct
(Heino 1998, Ripa and Lundberg 1996), and it should not
be confused with chaos or randomness.

Cybernetics, general systems theory, catastrophe the-
ory, and chaos theory all address deterministic dynamical
systems, systems where a set of equations determine how
a system moves through its state space from time t to time
t + 1. Another way of modeling complex behavior ex-
amines regularity that emerges from interaction of indi-
viduals connected together in complex adaptive systems
(CASs). The hallmark of this perspective is the notion
that at any level of analysis, order is an emergent property
of individual interactions at a lower level of aggregation.
Although there is no universally accepted paradigm for
describing CASs (Gell-Mann 1994b), four elements char-
acterize models that have particularly interesting impli-
cations for organization theorists.

Agents with Schemata. First, to model an outcome at
a particular level of analysis, one assumes that the out-
come is produced by a dynamical system comprised of
agents at a lower level of aggregation (Holland and Miller
1991). For example, in a model of an organization, agents
might be individuals, groups, or coalitions of groups.
Each agent’s behavior is dictated by a schema, a cognitive
structure that determines what action the agent takes at
time t, given its perception of the environment (at time t,
or at time t – k if theoretical considerations suggest ap-
plying a lag structure). Different agents may or may not
have different schemata (depending on one’ s theory), and
schemata may or may not evolve over time. Often,
agents’ schemata are modeled as a set of rules, but sche-
mata may be characterized in very flexible ways. For ex-
ample, an agent may select one rule from a suite of pos-
sible rules, or it may invoke fuzzy rules, or its cognitive
structure may be represented by a neural network (de-
scribed in more detail later in this article).

Self-Organizing Networks Sustained by Importing En-
ergy. Second, agents are partially connected to one an-
other, so that the behavior of a particular agent depends
on the behavior (or state) of some subset of all the agents
in the system. In systems dynamics models, variables are
connected to one another by feedback loops; in CAS
models, agents are connected to one another by feedback

ORGANIZATION SCIENCE/VOl. 10, No. 3, May-June 1999 219

This content downloaded from on Wed, 4 Sep 2013 16:13:49 PM
All use subject to JSTOR Terms and Conditions

PHILIP ANDERSON Complexity Theory and Organization Science

loops. Each agent observes and acts on local information
only, derived from those other agents to which it is con-
nected. Unlike cybernetic control theories, no single com-
ponent dictates the collective behavior of the system:
such systems self-organize (Drazin and Sandelands
1992). Maintaining a self-organized state requires im-
porting energy into the system (Prigogine and Stengers

Coevolution to the Edge of Chaos. Third, agents co-
evolve with one another. Each agent adapts to its envi-
ronment by striving to increase a payoff or fitness func-
tion over time (Holland and Miller 1991). Each
individual’s payoff function depends on choices that
other agents make, so each agent’s adaptive landscape-
mapping its behavior to its realized outcomes-is con-
stantly shifting (e.g., Levinthal 1997). The equilibrium
that results from such coevolution is dynamic, not static:
small changes in behavior at time t can produce small,
medium, or large changes in outcomes at time t + 1,
according to a power law (see Morel and Ramanujam,
this issue). Unlike chaotic equilibria, where small
changes in behavior frequently cause large changes in
outcomes, power-law equilibria lie at the edge of chaos
(Kauffman 1993).

Recombination and System Evolution. Fourth, com-
plex adaptive systems evolve over time through the entry,
exit, and transformation of agents. New agents may be
formed by recombining elements of previously successful
agents. Furthermore, the linkages between agents may
evolve over time, shifting the pattern of interconnections,
the strength of each connection, and its sign or functional
form. CASs can contain other complex adaptive systems,
as, for example, organisms have immune systems (Gell-
Mann 1994a).

CAS models represent a genuinely new way of sim-
plifying the complex, of encoding natural systems into
formal systems. Instead of making nonlinear systems
tractable by reducing them to a set of causal variables and
an error term, CAS models typically show how complex
outcomes flow from simple schemata and depend on the
way in which agents are interconnected. Rather than as-
suming that aggregate outcomes represent a homeostatic
equilibrium, they show how such outcomes evolve from
the efforts of agents to achieve higher fitness. By not forc-
ing scholars to understand all the parts of a complex sys-
tem in a holistic way, they allow investigators to focus
on an agent in its local environment. It becomes possible
to grasp complex behavior by varying assumptions about
the schemata, connections, fitness functions, or popula-
tion dynamics that characterize the agents. CAS models
afford exciting new opportunities for analyzing complex
systems without abstracting away their interdependencies

and nonlinear interactions. This is particularly important
for organizational scholars because interdependency is
central to modem conceptions of what an organization is
(Barnard 1938, Thompson 1967).

In the next section, I discuss each of these four features
in greater detail, examining how each contributes to or-
ganization theory. Then I turn to a discussion how these
ideas might lead to new ways of modeling organizational
phenomena, and conclude by assessing how complexity
theory creates new directions for research in strategic

Key Elements of Complex Adaptive
Systems Models
Agents with Schemata
Most conceptual and empirical models employed by sci-
entists studying organizations use a set of independent
variables to explain variation in one or more dependent
variables. Typically, outcomes at one level are explained
by causal drivers at the same level of analysis. CAS mod-
els take a different approach. They ask how changes in
the agents’ decision rules, the interconnections among
agents, or the fitness function that agents employ produce
different aggregate outcomes. These models are inher-
ently multilevel, because order is considered an emergent
property that depends on how lower-level behaviors are
aggregated. Accordingly, they respond well to contem-
porary calls for more integrative, cross-level research in
organization science (Rousseau and House 1994).

CAS models and ordinary causal models are comple-
ments, not rivals. It is not necessary for scholars to adopt
one or the other as the best way to analyze organizations.
Causal theories and tests that relate variables on the same
level identify important aggregate regularities and factors
that help create them. CAS models build on this foun-
dation, explaining observed regularities as the product of
structured, evolving interactions among lower-level units.
Good CAS models should not only explain established
findings, but successfully predict new aggregate regular-
ities and aggregate-level causal relationships.

Routinely, CAS models characterize agents as follow-
ing a set of rules (Gell-Mann 1994). Rule-based models
are also common in organization theory (Carley 1995),
but representing human actors in this way is problematic.
Institutional theorists have shown that rules are often ra-
tionalized myths (Meyer and Rowan 1977). Individual
goals and intentions may be only loosely related to be-
havior (March and Olsen 1976), and rules may well be
inferred from behavior instead of causing behavior
(Weick 1979). Scholars who view organizations as nat-
ural systems have shown that rules often do not govern

220 ORGANIZATION SCIENCE/VOl. 10, No. 3, May-June 1999

This content downloaded from on Wed, 4 Sep 2013 16:13:49 PM
All use subject to JSTOR Terms and Conditions



Complexity Theory and Organization Science



Organizations are mostly viewed as units that have a purpose, possess a structural form, and exhibit a given level of determinism and order. Complexity theory, in this instance is a pool of ideas revolving around the top bottom analysis approach used in understanding systems such as organizations, in fields such as strategic management. Its application areas comprise understanding how firms adapt to operational environments and how they handle uncertainty conditions. The theory treats firms and organizations as a collection of structures and strategies. The structure being complex that is, they are dynamic interaction networks, and they are adaptive meaning the collective behaviour change and organize themselves to fit the changes initiated by collection of events (Marion, 1999).

A theory of complex systems is important in unraveling the basic principles common to all systems. Presently there lacks a single integrated theory of complexity, but rather their exists theories that explain the common behaviors of a complex system such as:

Unification of themes, of a complex adaptive system (CAS), this is a system exhibiting behaviours such as learning, emergence, self organization, or co-evolution, which are popular across systems like human settlements or ant colonies. Appreciating these unification themes of CAS, helps to develop descriptions that relate to a case of an organization. The concept of self-organization is the ability of a system to instinctively self organizes itself into superior complex states, by interacting locally. This leads to renewal and reshaping of a whole system to adapt to external environment changes. Learning and adaptive behaviour is the capacity to learn and adapt to a complex system. The idea of an organization being complex and an adaptive system was derived in relation to the high levels of interconnectivity and technological advancements. Social systems, like organizations that are subsets exhibit a heap of complexity in their feature and form, by representing a complex interconnectivity web amid human beings who are capable of self-organization in order to respond to changes. However there is adaptation and learning involved at individual levels, system levels leading to development of direction and order, to empower groups into better handling of changes within its environment (Richardson, 2005).

In summary the notion of organizations being complex systems, capable of logically evolving strategies, processes, structures and self adjustment to changes in environment, imply novel roles and learning for managers as facilitators and guides for successful and transformative organizations.


Marion, R. (1999). The edge of organization: Chaos and complexity theories of formal social systems. Thousand Oaks, Calif: Sage Publ.

Richardson, K. A. (2005). Managing organizational complexity: Philosophy, theory and application. Greenwich, Conn: Information Age Publ.

Are you stuck with your online class?
Get help from our team of writers!