International Encyclopedia of Systems and Cybernetics

Selection is a limiting, negative agent that results in the elimination of some types (biological, economic or social). As observed by W.R. ASHBY, it is equivalent to a "mapping that reduces the original domain to a subset" (1964, p.96). It could however not work without the previous action of a creative variegating agent as was already observed in 1921 by A. BOGDANOV in his "Essays in Tektology", where he distinguishes between a conservative and a progressive selection (1980, p.74-85).

According to N.C. MARNEY and N.M. SMITH "in dynamic systems… the whole business of natural selection… amounts to the automatic generation of subsystems that are specially resistant to the perturbations characteristic of some particular subenvironment" (1964,p.119):

Such a mechanism must be some kind of branching self-organization process, whose precise nature and workings have remains quite elusive until the recent development of far-from-equilibrium thermodynamics. A promising possibility to model branching processes has been signaled by St. KAUFFMAN in his book "The Origins of Order: Self-Organization and Selection in Evolution" (1993).

Order may emerge from interactions between components, as shown for instance by causal processes, based on the effects of rules on some initial state, or in J. CONWAY's models known as "Game of Life", a particular case of self-organizing automata. According to KAUFFMAN, such processes may lead to the "edge of chaos", where the possibilities for innovation are maximal. This in turn opens the way to selection.

KAUFFMAN also hypothecizes that the general increase in complexity of systems (specially biological ones) progressively orients and restricts the role of selection. This could be a key to explain F. MEYER's evolutive acceleration (1954).

Complementary to this view is R. MARGALEF's: "Selection is a filter, but one that in turn evolves. There is a selection of mechanisms and motives for selection" (1980, p.93).