|
|
 |
|
|
|
|
tems "never get there." Improvement is usually much more important than optimization. When parts of the system do settle down to a local optimum, it is usually temporary, and those parts are almost always "dead," or uninteresting, if they remain at that equilibrium for an extended period. |
|
|
|
|
|
|
|
|
4. Complex adaptive systems anticipate. In seeking to adapt to changing circumstance, the parts develop "rules" (models) that anticipate the consequences of responses. At its simplest, this is a process not much different from Pavlovian conditioning. Even then, the effects are quite complex when large numbers of parts are being conditioned in different ways. The effects are still more profound when the anticipation involves lookahead toward more distant horizons. Moreover, aggregate behavior is sharply modified by anticipations, even when the anticipations are not realized. For example, the anticipation of an oil shortage can cause a sharp rise in oil prices, whether or not the shortage comes to pass. The effect of local anticipations on aggregate behavior is one of the aspects of complex adaptive systems we least understand. |
|
|
|
|
|
|
|
|
The objective of the Santa Fe Institute is to develop new approaches to the study of complex adaptive systems, particularly approaches that exploit interactions between computer simulation and mathematics. Computer simulation offers new ways of carrying out both realistic experiments, of flight-simulator precision, and well-defined gedanken experiments, of the kind that have played such an important role in the development of physics. For real complex adaptive systemseconomies, ecologies, brains, etc.these possibilities have been hard to come by because (1) the systems lose their major features when parts are isolated for study, (2) the systems are highly history dependent, so that it is difficult to make comparisons or tease out representative behavior, and (3) operation far from equilibrium or a global optimum is a regime not readily handled by standard methods. |
|
|
|
|
|
|
|
|
The Institute aims to exploit the new experimental possibilities offered by the simulation of complex adaptive systems, providing a much enriched version of the theory/experiment cycle for such systems. In conjunction with these simulations, the common kernel shared by complex adaptive systems suggests several possibilities for theory (cf. the work on the schema theory of genetic algorithms). In an area this complex, it is critical for theory to guide and inform the simulations, if they are not to degenerate into a process of "watching the pot boil." Theory is as necessary for sustained progress here as it is in modern experimental physics, which could not proceed outside the framework of theoretical physics. We need experiment to inform |
|
|
|
|
|