Carl Zimmer has written
that looks at software evolution from a biological point
"If the software performed better--in the sense that an organism had more reproductive success--the changes might become incorporated into the genome across an entire species. This was only a metaphor, but it was a powerful one. One example of its power is the rise of genetic algorithms. Rather than trying to find a perfect solution to a problem--the ideal shape for a plane, for example--genetic algorithms create simulations and tweak them through a process that mimics evolution. The algorithm can seek out good solutions very effectively. This sort of evolution resembles old-fashioned, closed-source software. All of the innovations happen in-house--that is, within a single species.
(Thanks to Martin Michlmayr.)
to post comments)