Dear Parasomnium,
I'm glad you are not claiming your solution is optimal, I thought I was going to have to put you in the bin for cranks making grandiose claims. As it is you haven't really done anything to make this issue clearer. The fact that you can generate some sub-optimal solution for the TSP using an evolutionary algorithm is still just an assertion, although not one I have any trouble believing. Just out of interest how many 'generations' did it take to 'evolve' your 99% good solution?
Why do you think something you yourself did, the details of which are known only to yourself, is a better basis for a discussion than research everyone can have equal access to? If you expect me to familiarise myself with classifications of P and NP problemks and the equivalency of NP optimal solutions why is it too much to expect you to familiarise yourself with the published research on digital evolution? This research should surely provide just as adequate a basis for this discussion as your own work?
Parasomnium writes:
# change in the outward appearance or behaviour of whatever is under scrutiny
# random mutation of whatever it is that (wholly or in part) determines the outward appearance of the thing under scrutiny
# non-random selection of more successful members of the 'population'
Your first point seems vague and probably redundant. What is your TSP solution other than what it is? What behaviour does it regulate? Similarly a sequence of RNA may evolve in and of itself. I understand how this relates to evolution but I think saying 'outward appearance' and 'behaviour' are misleading and give too much of an organismal slant when, after all, you could be talking about an improved binding affinity in some protein. I think 'some observable change' would be sufficient. Arguably the 'evolution' of a limb is only really the evolution of sequences of DNA such that the final form of the limb is altered, I don't think you can seperate these two levels out as you appear to be trying to do, perhaps if you showed how these levels equate to aspects of digital evolution your reasoning might be clearer. Your other two points simply seem to be random mutation and natural selection, which I have no argument with.
Parasmonium writes:
But if you look into the example of the Travelling Salesman problem, its solutions (itineraries!) certainly do not qualify as 'life' for any meaningful definition of life, yet one can use an evolutionary algorithm to arrive at a reasonable solution.
Demonstrate this, show how the itineraries, or better still of the code 'genomes' of digital organisms in Tierra or Avida, cannot qualify as living by any 'meaningful' definition but can still undergo the neccessary processes to qualify as 'evolving' using your criteria. Presumably at some point in this process you will finally have to actually give what you consider a 'meaningful' definition of life and we can begin to properly discuss this. At the moment this is simply a further bald reassertion of your claim.
TTFN,
WK
This message has been edited by Wounded King, 05-17-2004 04:58 AM