Abstrakt
An inspection of the full vocabulary of the words (16-mers) of the human genome reveals that the top of the list, ranked by word occurrence, contains almost exclusively simple repeats and words from Alu sequences - the most abundant dispersed elements. These excessive words can be considered ""generators"" and suggest a simple model of genome evolution: an everlasting intrusion of the generator sequences in the ""neutral"" regions of the genome and gradual mutational changes causing an increase in sequence complexity.
The way to detect the generators is to find those words all mutated forms of which appear less frequently. Examples of the generators are presented.