The invstigator and his colleagues organize a meeting at the Santa Fe Institute to study distributed autonomous systems. Such systems are comprised of many relatively simple elements that operate without central direction. Examples are metabolic systems, ant colonies, distributed computer architectures, and the immune system. Emphasis is put on the immune system, because of its great intrinsic interest, its importance in health and disease, and the massive (but incomplete) detail known concerning the operation of its trillions of cells. Leading theorists and experimentalists in immunology present a distillation of present day experimental knowledge together with mathematical and computer models based on this knowledge. These experts seek to formulate general principles that apply not only to the immune system but also to the other distributed autonomous systems that are under study. Experts on the other systems report on their specialties and also try to distill principles that are broadly applicable. Half the participants are graduate students and postdocs, who can adopt the novel views expressed in their future research. This meeting is concerned with the study of systems that can accomplish a variety of complex tasks with a large leaderless group of relatively primitive components. The major example to be studied is the immune system, with its trillions of cells and hundreds of signaling chemicals. Experimentalists and theorists report on advances in understanding how the immune system can so expertly perform a spectrum of tasks such as killing the large variety of bacteria and viruses that invade the body. Other experts report on advances in understanding other complex systems, such as the metabolic system that turns our food intake into energy and growth, or systems of fictitious "agents" that work in novel computer programs that attempt to reproduce aspects of creative thought. The clash of ideas that come from simultaneous consideration of these different systems should generate principles of wide applicability. Such principles are essential if we are to advance beyond the stage of collecting pieces of data that add to our understanding of elements of system behavior but leave us more and more baffled as to how all these elements combine to give effective system performance, and how we could ever manipulate system behavior with any confidence. In time, identification and validation of these principles can have far-reaching applications ranging from aid to physicians in generating manipulations that help the immune system fight disease, to enabling computer programmers to find novel ways to fight "virus" and other intrusions into computer systems and to design computer systems that "think" less rigidly and more creatively than present "intelligent" computers.
