International Encyclopedia of Systems and Cybernetics

A statement about some future state of a system.

Scientific prediction in the classical sense is limited in practice. It is possible only within the frame of reference of precise and limitative conditions.

T. BAUMGARTNER et al. give the following example: "It must be realized that the pinnacle of prediction occurs in physical mechanics and is based upon the concept of the isolated point mass, subject to forces acting at a point. These notions and the basic conservation laws permit derivation of the equations of "point mass dynamics", which can be extended to arbitrary bodies, analyzed as finite collections of point masses, through consideration of the "continuity of matter" and calculus. The precision with which astronomical trajectories are determined and the guidance of spacecraft to within yards of a predetermined landing site provide striking validation of the models and the principles upon which they are based" (1976, p.34).

There are however many situations, even in physical sciences, where these postulates lead to somewhat, or grossly erroneous predictions (see for example POINCARÉ's 3-bodies problem) or become downright useless in practice (as in LORENZ chaotic meteorological dynamics).

Predictions are thus heavily dependent on the alternative models selected to make them. As data relative to processes can be in many cases accomodated (or fudged!) to different models (for ex. linear, logistic, asymptotic, exponential growth), specially when they cover too short a span of time, the prediction may very well be implicitly stated within the model.

Moreover, biological and social systems offer such high degrees of complexity that prediction must cover not one, but an array of interconnected processes. To be more or less possible, prediction demand then quite more intricated methods.