Munich Personal RePEc Archive

Transfers by force and deception lead to stability in an evolutionary learning process when controlled by net profit but not by turnover

Friedrich, Thomas (2019): Transfers by force and deception lead to stability in an evolutionary learning process when controlled by net profit but not by turnover.

This is the latest version of this item.

[img]
Preview
PDF
MPRA_paper_97806.pdf

Download (656kB) | Preview

Abstract

An evolutionary process is characterized by heritable variation through random mutation, positive selection of the fittest, and random genetic drift. A learning process can be similarly organized and does not need insight or understanding. Instructions are changed randomly, evaluated, and better instructions are propagated. While evolution of an enzyme or a company is a long-lasting process (change of hardware) learning is a fast process (change of software). In my model the basic ensemble consists of a source and a sink. Both have saturating benefit functions (b) and linear cost functions (c). In cost domination (b-c<0) source gives substrate and in benefit domination (b-c>0) sink takes it - both at free will - thus creating a basic superadditivity. It is not reasonable to give when b-c>0 or take when b-c<0. However, with force and deception source and sink of an ensemble can be overcome to give or take although it is not reasonable for them. This leads to further superadditivity within the ensemble. But now subadditivity will appear in addition in certain regions of the transfer space. I observe organisms or companies learning by trial and error to optimize superadditivity without changing the characteristics of the benefit function or the cost function. The role of a third-party master of an ensemble to create superadditivity in the absence of cost domination in source or benefit domination in sink by force and deception is investigated in connected and unconnected ensembles. Employees and companies can be rated according to turnover or net profit. My model confirms the superiority of the benchmark net profit as self-limiting, sustainable incentive in an evolutionary learning process.

Available Versions of this Item

UB_LMU-Logo
MPRA is a RePEc service hosted by
the Munich University Library in Germany.