|
|
|
|
|
|
|
plinary consortium, the Santa Fe Institute, dedicated to the study of complex adaptive systems. The Santa Fe Institute, by providing a focus for intensive interactions among its collection of Nobel Laureates, MacArthur Fellows, Old and Young Turks, and bright young postdocs, has already made a substantial impact in the field of economics. Current work emanating from the Institute promises similar effects in fields ranging from studies of the immune system to studies of new approaches to cognitive science. The future for studies of adaptive systems looks bright indeed. |
|
|
|
|
|
|
|
|
Fifteen years should provide perspective and a certain detachment. Despite that, or because of it, I still find the 1975 preface surprisingly relevant. About the only change I would make would be to put more emphasis on improvement and less on optimization. Work on the more complex adaptive systemsecologies, for examplehas convinced me that their behavior is not well described by the trajectories around global optima. Even when a relevant global optimum can be defined, the system is typically so "far away" from that optimum that basins of attraction, fixed points, and the other apparatus used in studying optima tell little about the system's behavior. Instead, competition between components of the system, aimed at "getting an edge" over neighboring competitors, determines the aggregate behavior. In all other respects, I would hold to the points made in the earlier preface. |
|
|
|
|
|
|
|
|
There are changes in emphasis reflected by two changes in terminology since 1975. Soon after the book was published, doctoral students in Ann Arbor began using the term genetic algorithm in place of genetic plan,emphasizing the centrality of computation in defining and implementing the plans. More recently, I've advocated implicit parallelism over intrinsic parallelism to distinguish the "implicit" workings of the algorithm, via schemata, from the parallel processing of the populations used by the algorithm. |
|
|
|
|
|
|
|
|
As a way of detailing some more recent ideas and research, I've added a new chapter, chapter 10, to this edition. In part, this chapter concerns itself with further work on the advanced questions posed in section 9.3 of the previous edition. Questions concerning the design of systems that build experience-based, hierarchical models of their environments are addressed in section 10.1 of the new chapter. Questions concerning speciation and the evolution of ecologies are addressed in terms of the Echo models in section 10.3. The Echo models, besides being concerned with computer-based gedanken experiments on these questions, have a broader purpose. They are designed to facilitate investigation of mechanisms, such as competition and trading, found in a wide range of complex adaptive systems. In addition to these discussions, the new chapter also includes, in section 10.2, some corrections to the original |
|
|
|
|
|