OT: Evolution

[Simon Bowring]

On Tue, 19 Sep 2000 06:14:04 -0700, Bill Wood wrote:

>Evolution is change. Usually random, but not always. Then the
>change, whatever its origin, is thrown into the arena of real life
>and it either survives or it doesn't. Etc.
Agreed!

>Now that we are getting visibility of human DNA coding, it seems to
>be a ... mess ... not even rudimentary error correction
I'd dispute that, but not backed up with facts right now, but even
the number of bases (4) seems like it may be an evolutionary 
adaptation which is a compromise between reliability and information 
content (apparently)!

>This supports the notion of a random process. This is probably not a good 
>model for software development, but it may be better than what is 
>actually done.

ie. Genetic alrorithms - genetic algorithms work just fine provided 
the "test for fitness" test is "correct". For email programs, one of 
the tests for fitness would be to pass a complete suite of standards 
compliance tests ;-)

A few years ago, someone in BT's labs came up with a previously 
unknown algorthm for "solving" (i.e. producing good, but not 
necessarilly the best solution) the travelling salesman problem 
using these techniques.

Simon