|
|
|
|
|
|
|
brought to bearthe gedanken experiment. As the name implies, a gedanken experiment is a thought experiment. It extracts a few elements from a process in order to examine, logically, some critical effect produced by the interaction of these elements. Computers offer a way of extending the scope of gedanken experiments to much more complex situations. Echo has been designed as a computational base for gedanken experiments on complex adaptive systems. |
|
|
|
|
|
|
|
|
Echo, and other models of complex adaptive systems, are readily designed for direct simulation on massively parallel computers. It is also possible to design interactive interfaces for these simulations that permit ready, intuitive interactions with the ongoing simulation, much as is the case with flight simulators. Thus, the "logic" of the simulation can be combined with the human's intuition and superb pattern recognition ability to provide quick detection of interesting patterns or events. This has the double value of providing reality checks on the design, while allowing investigators to bring their scientific taste and intuitions to bear in creating and exploring unusual variants. |
|
|
|
|
|
|
|
|
By looking for pervasive phenomena in these gedanken experiments, we can study complex adaptive systems with a new version of the classic hypothesize-test-revise cycle. The "test" part of this cycle is particularly important because complex adaptive systems, as mentioned earlier, typically operate far from a steady state. They are continually undergoing revisions, and their evolution is highly history dependent. This, combined with the nonadditive nature of the internal interactions, makes it difficult to do controlled experiments with real complex systems. Computer-based gedanken experiments should help fill the gap. |
|
|
|
|
|
|
|
|
In examining complex adaptive systems, there is one property that is particularly hard to examine in situ. Complex adaptive systems form and use internal models to anticipate the future, basing current actions on expected outcomes. A system with an internal model can look ahead to the future consequences of current actions without actually committing itself to those actions. In particular, the system can avoid acts that would set it irretrievably down some road to future disaster ("stepping over a cliff"). More sophisticated uses of an internal model allow the system to select current "stage-setting" actions that set up later advantageous situations (as in Samuel's [1959] use of "lookahead"). As pointed up earlier, the very essence of attaining a competive advantage, whether it be in chess or economics, is the discovery and execution of "stage-setting" moves. Internal models distinguish complex adaptive systems from other kinds of complex systems; they also make the emergent behavior of complex adaptive systems intricate and difficult to understand. |
|
|
|
|
|
|
|
|
Internal models offer a second advantage in addition to the advantage of |
|
|
|
|
|