TOP DOWN
| Collection | International Encyclopedia of Systems and Cybernetics |
|---|---|
| Year | 2004 |
| Vol. (num.) | 2(2) |
| ID | ◀ 3587 ▶ |
| Object type | General information, Methodology or model |
The organizational mode of a machine made of subsystems performing subfunctions and ordered in such a way as to contribute to the accomplishment of the predefined global function of the machine.
A good example is the building of a plane, for which a very detailed blueprint has been previously established, describing in great detail the interconected functions of the multiple parts, as well as the parts themselves.
Such a system is a strongly integrated unit. So it is because it has been imaginated and constructed in a preconceived way by its creator. It is quite well adapted to the finality assigned to it, but is unable to adapt itself to different situations, outside of the limits of his predefined finality.
In classical A.I., the top-down approach implies the need for algorithmic rules, which, by nature are not adapted to not foreseable situations.
The “top-down” organizational mode is complementary but also antinomic to the “bottom-up” one, which is better adapted to social type situations. (See more explanations under the heading “Intelligence (Distributed Artificial)”.
The “top-down” models fit centralized control of interactions, but tend to ignore the real complexity of systems. The “bottom up” models tend to become over-complicated, very difficult to integrate in a coherent way and to apply to real situations.
Bottom-up construction and top-down management , in natural systems, regulation or control are in a dialectical relation.
Any process of emergent complexity implies recurrent interactions , that must be multi-correlated and stabilized.
If taken in a properly dynamic and sequential sense, H. SABELLI's Principle about “Priority of the simple , supremacy of the complex ” resumes very well the unavoidable complementarity of the two processes.