🚀 go-pugleaf

Re Genetic Algorithms

I've been doing some work in this arena, it is very promising. It can also drive
you nuts. Some of the problems:

They are fairly easy to implement within a limited domain, but get nasty when
you expand the horizon (ie. add terrain considerations, for example).

Making individual components smarter isn't too bad, getting them to cooperate is
trickier.

Finding an evaluation algorithm can be tricky.

Fun stuff to work with though. (I'm currently writing a GA to predict winners at
the local dog track, how's that for putting my money where my mouth is?)

--
Steve Bittrolff

The previous is my opinion, and is shared by any reasonably intelligent person.