next up previous contents
Next: Self-Reproduction and Evolution Revisited Up: Problems with Tierra-like Models Previous: No Competition for Matter

Evolving a Self-Reproduction Algorithm

Another concern with the Tierra approach is the fact that the individual organisms are encodings of self-reproduction algorithms. This is related to the concern over imposing a predefined structure on the organisms, but is a more fundamental worry. The fact that the copying process is explicitly encoded in the individual organisms in Tierra-like systems, rather than being implicit in the physical rules of the system, might have serious consequences for the evolvability of the organisms. Additionally, Tierran programs do not appear, at least at first sight, to have a clear genotype-phenotype distinction, and this may also hinder their evolvability.

These issues are a source of concern, but I have not yet offered any detailed analysis to suggest how great a problem they actually represent, if indeed they represent a problem at all. As these issues relating to self-reproduction are fundamental to the goal of creating an open-ended evolutionary system, they deserve a more thorough analysis. Such an analysis is provided in the following section.

Tim Taylor