International Encyclopedia of Systems and Cybernetics

The possibility for an observer to reduce the complexity of the real world by constructing algorithms which produce an ordered and simplified understanding of a number of phenomena.

J. BARROWS explains: "If it is possible to store the information in an abbreviated form shorter then the sequence itself, then the sequence is regarded as nonrandom and we call it compressible". If no such abbreviation exists, then the sequence admits no representation other than the complete explicit print-out of itself. In this case, we call it "incompressible".

"… this notion of compressibility gives simple ways of characterising many of our intellectual activities. We recognize a possible new definition of "science" as being simply the search for compressions: the laws of Nature are the compressions of our sense data. The discovery of a "Theory of Everything" would be the ultimate compression. Moreover, the apparent success of this process hinges upon two superficial features of things: the physical world that we observe seems to be surprisingly amenable to compression, and the brain is remarquably good at efecting compressions when presented with events.

"Clearly, this compressibility and the brain's remarquable ability to make sense of complicated things is an important necessary condition for our own existence. We could not survive as "intelligent" observers… in a world where no compressions were possible, or with brains that produced imaginary or erroneous compressions. A certain level of predictability and innate predictive power is required for the successful evolution and survival of living things" (1991, p.43).

Of course, as noted by BARROWS our brain can perfectly produce these imaginary or erroneous compressions, as for example "see canals on Mars and all manner of exotic things lurking into inkblots". Thus, let us use algorithms, but let us also beware of them!

Cybernetic and systemic concepts offer however new tools for compression.

G. KLIR, for instance, proposes his activity arrays as compressed behavior generating devices, allowing eventual law-discovery processes (1975, p.167).

→ Algorithmic incompressibility