Re: Announcing Breeder 1.0

[Ludo Pagie]

Hi all,

After a very quick reading of the manual of Breeder 1.0 it seems to me that the
library does not allow for any structure in the (evolving) population other
than differences in genetic material.
Being a lurker of type 2 I don't have a clear idea yet if it would be possible
to have a buch of agents running around, each having one genome.
IMHO, the agents should behave (partially) according to their genome, survival
and mating ('fitness') would thus (partially) depend on their genome. Mating
should involve genetic alteration (mutation, recombination) of the genomes,
leading to an evolving population of agents.
The main point; the 'fitness' of agents should depend on their behaviour, in
their surroundings, competing with their neighbours. Not on some predefined
static fitness function.
You guessed that I'm not talking about the optimization of anything, I'm
talking about evolution of (something).
What I understand of swarm and Breeder sofar, it should be possible to have
population of genomes in single agents, where in each agent some sort of
optimal behaviour is evolved. However, where is the (static) fitness function
coming from? Do we know in advance what the optimal ...whatever... should be?
Isn't an important point in this sort of individual oriented modelling that
knowledge of individual behavioural strategies does not lead to knowledge of
how the population will behave, and thus what a good/optimal individual
strategy is if you are surrounded by such a population? (I'm drifting, I know)

Coming back to my question; is such use of the Breeder lib possible? Somebody
else working on it (any GP variants?).

Just some thoughts, Ludo.