Stuart Kauffman
Paper #: 90-020
The general problem which arises in investigating the capacities of complex systems to adapt lies in understanding both the “functional” and the “dynamical” order which integrates these systems. E. coli “knows” its world. A wealth of molecular signals pass between a bacterium and its environment, which includes other microorganisms, and higher organisms. The signals entering the bacterium are harnessed to its metabolism and internal transformations such that, typically, the cell maintains itself, replicates, and passes its organized processes foward into history. Similarly, a colony of E. coli integrates its behavior. The organisms of a stable ecosystem form a functional while the niches occupied by each jointly add up to a meshwork in which all fundamental requirements for joint persistence are met. Similar features are found in an economic system. The sets of goods and services comprising an economy form linked mechwork of transformations. The economic niches occupied by each allow each to earn a living and jointly add up to a web in which all mutually defined requirements are jointly met. Both biological and technological evolution consist in the invention of profoundly, or slightly novel organisms, goods or services which integrate into the ecological or economic mesh, and thereby transform it. Yet at almost all stages, the web retains a kind of functional coherence. Furthermore, the very structure and connections among the entities sets the stage for the transformation of the web. In an ecosystem or economic system, the very interactions and couplings among the organisms, or goods and services create the conditions and niches into which new organisms, or goods and services can integrate. The “web” governs its own possibilities of transformation. Similar functional integration of roles, obligations, and institutions apply at societal levels. The revolution occurring in East Europe and the U.S.S.R. in the “anni mirabili,” are accompanied by a sense that the Soviet system is an integrated whole with the property that if one, or a few features are removed or altered, the entire system must transform to something quite different--and whole. In June 1989 the Communist leaders in China saw fit to kill their students in Tienamin Square. Why those leaders did so is clear: The students were demonstrating for increased democracy. Their government feared the conssequences would transform Chinese Communism. In short, the puzzle is not to understand what China's leaders did, but rather to understand what they “knew.” In a real and deep sense the Chinese government knew that, were a few features of their system altered, the entire edifice stood in danger of dramatic transformation. What, indeed, did they know? In the biological and social sciences we badly lack a body of theory, indeed even a means of addressing these issues: What is a functional whole and how does it transform when its components are altered? In this article I shall develop an outline for a fresh approach to these important issues. The approach is based on the use of random grammars. The objects of the theory are string of symbols which may stand for chemicals, goods and services, or roles in a cultural setting. Symbol strings act on one another, according to the grammar, to yield the same or other symbol strings. Thus, the grammar specifies indirectly the functional connections among the symbol strings. It defines which sets strings, acting on other sets of strings, produce which sets of output strings. These mappings are the functional couplings among chemicals in a protoorganism, among a population of organisms in an ecosystem, and become the production technologies in an economy. Diverse grammars model diverse possible chemistries, or possible production technologies. By studying the robust features of functionally integrated systems which arise for many grammars that should fall into a few broad “grammar regimes,” it should be possible to build towards a new theory of integration and transformation in biological and social sciences. Among the features we will find are phase trasitions between finite and potentially infinite growth in the diversity of symbol string in such systems. This phase transition may well underlie the origin of life as a phase transition in sufficiently complex sets of catalytic polymers, and a similar phase transition may underlie “take off” in economic systems once they attain a critical complexity of goods and services which allows the set of new economic niches to explode supracritically. My suggestion to study random grammars grows from work initiated by myself, in investigating “autocatalytic sets of polymers,” thereafter carried on in collaboration with Doyne Farmer, Richard Bagley, Norman Packard, and Walter Fontana. In particular, I believe the recent extensions by Walter Fontana concerning autocatalytic sets are exciting and important. All point towards a new way to investigate the emergence of functional integration and adaptation in complex systems.