International Encyclopedia of Systems and Cybernetics

Statistics on probabilities are relative to the past. They reflect the numerous and varied results of a random process that already has taken place, and thus they are not any more probabilities. They are a rationalization of historical, unchangeable evidence, even if they provide us with at least a measure of information about the nature of the process.

P. VENDRYES clarify this point as follows: "When the same random process is repeated a great number of times, each time requires independence of the repetitions. The relative frequencies of the various possible outcomes tend towards the respective probabilities of each outcome. This tendency increases with the number of repetitions. These frequencies endow the probabilities with realistic values. They have been referred to as statistical probabilities. But these frequencies were actually manifest in the past, and belong to this past, whereas the probabilities are virtual and are turned toward the future" (1989, p.148).

Synthetically, frequencies are historical while probabilities are conjectural, i.e. based on some hidden postulates. The result, as stated by E. JANTSCH is that: "Most of the models applied so far assume equilibrium conditions and define probability accordingly", while "The recognition of nonequilibrium conditions… leads to vastly different concepts" (1975, p.198).