300 years ago, Newton, Laplace, and Descartes introduced a revolution in scientific thinking: with sufficiently precise understanding of initial conditions, the future could be predicted by applying a simple set of natural laws. Although this "clockwork universe" paradigm has long since been discredited by physicists, its intellectual descendants still haunt our institutions, bureaucracies, economies, universities, software development methodologies, and general zeitgeist. If only our policies and procedures, economic measures, university education, software metrics, and social indicators were more precise, we could apply a few simple "natural laws" and predict the future....
The scientific revolution of the past 300 years has undeniably yielded many tangible benefits. However, from many viewpoints, the clockwork universe mindset is proving insufficient to meet the real-world needs and expectations. We have learned from chaos theory that we cannot simple collect ever more precise initial conditions to predict the behavior of a system. We have learned from software engineering efforts that no matter how hard we try to establish the perfect initial conditions for software development (requirements) that we cannot generate a smooth, laminar flow software development process.
There are many indicators which indicate that the reductionistic frenzy which has driven the lust for precision in our times is dissipating. Chaos theory, non linear dynamics, complex adaptive systems, genetic algorithms, artificial life, virtual reality, complexity theory, simulation, emergent computation, and fuzzy set theory are but a few of the disciplines affecting this. Rather than seeing the world as a set of complex initial conditions and simple laws, this new world view sees the things in a much more adaptive, "organic" manner. Systems begin with a relatively simple initial condition: a primordial soup, so to speak. They then evolve over time according to rules of selection and fitness. The world is not necessarily so deterministic. Positive feedback loops create unpredictable emergent properties. Autocatalytic organizations emerge which appear to violate the rules of entropy: they spontaneously exhibit increasing complexity.
This paper discusses the problem of managing a large scale enterprise-wide information system. It builds on the technology embedded in the World Wide Web (WWW), complex adaptive systems, and object-oriented technologies to provide a framework within which an enterprise may construct an adaptive, evolutionary system which meets its needs in a cost effective manner.
The traditional "clockwork universe" view of the world which was advanced by Newton and Laplace was based on the execution of a few simple laws (e.g., F=MA), starting from complex initial conditions (the position and momenta of all particles in the universe). Although physicists gave up on this idea at the time of Heisenberg, its analog has carried forth in many other areas of thought. This "divide and conquer" mindset has lead us to divide and conquere organizations in economics, other sciences, education, medicine, and information technologies, as well as other places.
Computer science has been subject to this kind of thinking, which has resulted in many different levels of reductionistic thinking: hierarchical decomposition, structured analysis and programming, various systems engineering approaches, etc. These approaches work in some domains which can be characterized as having complex initial conditions (expressed as requirements), executed with simple laws (the variants of structured programming). We "normalize" our data by structuring it into densely packed rectangular tables; data which doesn’t fit these rigid conditions are considered "unnormalized", which is somewhat akin to the pope calling all people not in his church "non-catholics." Indeed, the prefix "hier" means sacred or holy. Hierarchical decomposition of systems into functional components has become the sacred quest of today’s computer systems architects.
The model presented in this paper presents a radically different view of the information system. It begins with simple initial conditions: as simple as possible, and then allows the system to evolve as a complex adaptive system which is adapting to its environment. Holland (1) describes these classes of systems: