This is a Python implementation of the Dissociated Press algorithm, which
generates text based on another text.
The principle is fairly simple, at least if my interpretation is correct: You
choose any N consecutive words from the reference body of text, and add those.
Then you look for all occurrences in the text of those N consecutive words,
randomly choose one, and add the N words that follow. Then you repeat this
iteratively.

The Jargon file entry provides a decent working definition:
http://www.catb.org/jargon/html/D/Dissociated-Press.html