Dr. Norman L. Johnson Contact norman@santafe.edu Documents on the Science of Diversity Return to Home Page Check here in the future for new documents

Summarizes in 10 pages how decentralized, adaptive systems evolve by a developmental process. How each developmental stage has different mechanisms, from competition in the Immature/Formative stage to emergent collective effects in the Mature/Co-Operational stage to exclusive cooperation in the Senescent/Condense stage. Diversity is the key to understanding how each stage improves and the system is vulnerable (or not) to failure within a changing environment. An expanded version of this paper will appear soon.

This paper was presented to a group of social scientists (anthropology, political science, economics, business, etc.) concerned with how decentralization influences the functioning of organizations. This paper contains careful definitions of terms, such as agent, local/global, robust/fragile, self-organizing, chaotic/stable, diversity and common world view (context). The advantages and disadvantages of decentralized systems are presented. The abstract follows:

Research on how groups can solve problems better than experts has suggested a general understanding of the how diversity plays an important role in a variety of decentralized systems: ecosystems, social groups, large organizations, political systems, free market economies, the stock market, our society - any system where there are individuals or groups make decisions or solve problems without total centralized control or planning. If fact, it appears that diversity is often important in systems thought to be homogeneous and centralized. Both decentralization and diversity are necessary attributes to achieve the self-organizing properties of interest. Three important observations result from the study that should be helpful to researchers modeling decentralized systems. The first is that dominant processes in these systems can change over time (a developmental view of systems). The second is that self-organization without competition or cooperation is overlooked as an important process, compared to the well-studied actions of competition and cooperation. This type of self-organization can be viewed as a mechanism for achieving higher performance, robustness and conflict resolution without invoking the mechanisms of competition or cooperation for the agents in the system. And finally, measures of robustness of the system also need to be evaluated, in comparison to the often studied measures of performance and efficiency. A consideration of robustness clarifies the role of diversity.

This paper presents the behavior of a model problem that expresses the improvement of groups by processes that do not involve the tradition methods of selection, cooperation or competition. It shows how groups using this emergent capability can perform better than the average individual. This paper captures in detail most of the technical results that are discussed in other document at this site.

The following two papers illustrate the same importance of diversity from an economic viewpoint. Particularly they present the surprising result that a collective actually does worse if only the high performing individuals are included in the collective process. While many of the conclusions are the same as observed in the simulations in Collective Problem Solving: Functionality beyond the Individual, the model problem is quite different. This suggest the conclusions are not specific to a certain model.