Research and Advances
Artificial Intelligence and Machine Learning Adaptive complex enterprises

Introduction

Posted
  1. Introduction
  2. Complex Adaptive Systems
  3. Information-Theoretic Interpretation of Complexity
  4. IT in a Complex Enterprise
  5. Strategic Planning in ACE
  6. References
  7. Author
  8. Sidebar: Defining Terminology

It is common knowledge that individuals, the communities in which we live, and the organizations we create adapt over time and changing conditions. It is also generally understood that organizations, communities, and individuals are all complex entities. To state that organizations are adaptive complex enterprises (ACE) is neither novel nor new: What is new, however, is that we now have a language and a set of constructs that allow us to formally conceptualize and discuss these concepts. (See the sidebar for a brief discussion of the terms adaptive complex enterprises and complex adaptive systems.)

Heisenberg’s uncertainty principle shook the foundations of Newton’s simple clockwork world. To Heisenberg, his new world "appears as a complicated tissue of events, in which connections of different kinds alternate or overlap or combine and thereby determine the texture of the whole" [3]. Quantum theory further complicated the physicists’ world. Biologists and other natural scientists also discovered that they were studying complex phenomena that demonstrated emergent and adaptive behaviors. Natural and physical scientists soon found themselves in search of a philosophy of science different from the one that seemed to have served them so well for almost 300 years.


Most nontrivial naturally occurring processes are complex. They have managed to survive and thrive by being adaptable.


The origins of the modern formal treatment of complexity can be traced to René Thom’s mathematical treatise on catastrophe theory [6]. In describing seven types of mathematical singularities, he provided a classification of catastrophes, which was a static description of morphogenesis, not its dynamics. The dynamics of abrupt change came from a different context in the form of chaos theory and the study of deterministic nonlinear dynamical systems. Chaos theory focused attention on systems in which minor perturbations lead to major consequences such as the origins of a hurricane being famously attributed to the fluttering of a butterfly’s wings in a distant rain forest.

Complexity studies, particularly those associated with the Santa Fe Institute, combine the interest of catastrophe theory in discontinuous change with the dynamics of nonlinear systems, which is the focus of chaos theory. Complexity exists at the "edge of chaos," somewhere between too much and too little order.

Most nontrivial naturally occurring processes are complex. They have managed to survive and thrive by being adaptable, though it is not clear whether this adaptability is gradual as Darwin suggested or consists of a series of punctuated equilibria, as proposed by the Gould-Lewontin hypothesis [2]. In efforts to mimic such adaptability, artificial systems, in the sense of Simon [5], as being those with human origins, have been made adaptable by the introduction of feedback loops [1]. Simple adaptive systems, such as thermostats, are designed to detect differences. When this difference between a previous state and the current state is determined to be significant, there is an adaptive response. In these situations, negative feedback results in a corrective action, which returns the governed system to its previous state. Returning a system to stability is not always the desired objective. In fact, the realization that positive feedback, which amplifies the instability, can be a force for change has spawned an interest in identifying adaptive responses for seeking new and potentially superior, equilibria.

Social and management scientists have begun to draw upon these ideas concerning complexity and adaptability to make sense of their own worlds. The focus of this section is on adaptability in systems and processes that humans have created and now attempt to manage and, perhaps, control.

Back to Top

Complex Adaptive Systems

From a manager’s perspective, in order to give control to those in charge, the task is to design adaptive systems or processes that bring back or push forward conditions that are far from equilibrium into some state of equilibrium. A new realization appears to be emerging that in dealing with complex situations, we need flexibility and adaptability such that systems or processes are not frozen because they are too tightly constrained nor are they dysfunctional such that they disintegrate due to too little order.

In the first article in this section, Tan, Wen, and Awad introduce the traditional notion of a complex adaptive system [4] and discuss how a health care system can be better managed when viewed as such. In particular, they discuss the utility of feedback loops to identify levels of stability of the system. This feedback provides managers with information for determining how far or near the system is to stability and thereby develop the appropriate adaptive response.

Back to Top

Information-Theoretic Interpretation of Complexity

Efforts to measure complexity have led researchers to conclude that complexity itself is a complex concept. When the source of complexity is seen in terms of unpredictability or uncertainty, probability is an obvious measure of the level of uncertainty. The second law of thermodynamics and the associated measure of entropy can be used as another indicator of the complexity of a system. Shannon’s measure of information, often interpreted as the negative of entropy, provides yet another measure. Hence depending upon how we choose to define it, related, yet different measures may be used to measure complexity. These, however, are primarily measures of orderliness or physical complexity. In "Test Beds for Complex Systems," Jones and Deshmukh point out that we do not have good measures of uncertainty or complexity in terms of meaning or sense making.

Value chains exhibit local as well as global complexity. The decision making that occurs at the local level is very different from that occurring at the global level and hence the complexity as well as its measure are different. Jones and Deshmukh distinguish between static and dynamic complexity and make the case that information, as the raw material for decision making, is vital for good performance. However, they contend that we need to learn a lot more about the relationship between information, uncertainty, complexity, and the performance of physical systems, and suggest test beds that would help us make progress toward that understanding.

Back to Top

IT in a Complex Enterprise

Information technology plays an important role in developing the test beds described by Jones and Deshmukh. While IT has been very successful at automating routine tasks, its record of accomplishment with nonroutine tasks and adaptability has been quite poor.

Ramanathan suggests that the recent increase in processing capacity and speed enables IT to react in real time. Hence, rather than using IT to control and minimize variation, these technologies can now be used to embrace variability: IT is providing multiple benefits in managing complex situations. Ramanathan provides yet another characterization of a complex enterprise and offers four different variability patterns as sources of complexity. In her characterization, the adaptive complex enterprise’s behavior is the result of agents that can self-organize into an evolving process structure capable of executing an effective sequence of decisions to respond to incoming requests. She argues that although the resources used at different levels of an organization may be different—as Jones and Deshmukh suggest—information may be the chief resource at the upper management levels and physical resources are utilized at the operational level, the underlying processes are the same. Regardless of the level, the tasks can be seen in terms of identifying requirements, executing an operation, or delivering on a request. Their fractal nature is revealed when a detailed look at these tasks yields the same three components.

Back to Top

Strategic Planning in ACE

In the final article, Ramnath and Landsbergen discuss the implementation of some of these ideas as they report the lessons learned from an ongoing effort to implement a strategic plan for a U.S. city government IT department. An objective of this effort is to develop and implement a unified organizational and IT plan to establish adaptive systems that can sense and respond effectively to routine and nonroutine requests.

It is likely that most readers have not had any formal training in the holistic study of the topics discussed here. The articles illustrate that the authors’ own thinking regarding the notions of adaptability and complexity are still evolving. The hope is that although the articles that follow shed light on limited areas, there will be enough variety and sufficient overlap that eventually the whole field, however characterized, will be better illuminated.

Back to Top

Back to Top

Back to Top

    1. Bateson, G. Steps to an Ecology of Mind. Ballantine Books, New York, 1972.

    2. Gould, S.J. and Lewontin, R.C. The spandrels of San Marco and the Panglossian paradigm: A critique of the adaptationist programme. In Proceedings of the Royal Society of London B 205 (1979) 581—598.

    3. Heisenberg, W. Physics and Philosophy. Harper Torchbooks, New York, 1958.

    4. Holland, J.H. Hidden Order: How Adaptation Builds Complexity. Perseus, Reading, MA, 1995.

    5. Simon, H.A. The Sciences of the Artificial, 3rd edition. MIT Press, Cambridge, MA, 1997.

    6. Thom, R. Structural Stability and Morphogenesis: An Outline of a General Theory of Models. Westview Press, 1972.

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More