Beyond Digital Naturalism: The Need for Clear Goals^7.21

As I suggested in Section 3.3.1, Tierra and similar systems can be viewed as an exploratory investigation into the potential of synthetic evolutionary techniques. Such exploratory studies are a useful and normal aspect of any experimental science. However, as the reappraisal presented in Section 7.1.1 of this chapter suggests, a more rigorous methodology must be adopted if the approach is to be of any further scientific value. In particular, artificial life systems should be designed to address specific issues; these issues may be big or small, but they must be specific. Associated with this is the need for a coherent theoretical framework upon which the system should be grounded. This framework serves as a proposal for an explanation (at some level) of the phenomenon being addressed,^7.22 and explicitly states which aspects of the system's design are claimed to be relevant to the phenomenon. The system should, to the greatest extent practicable, model these aspects and nothing else; in other words, it should ideally be a minimal model. In practice there will always be choices to be made when deciding how to model a given object or process (cf. David Marr's work, mentioned in Section 3.1.3), but the theory should claim that such choices will be irrelevant with respect to the particular phenomena being investigated.

With this need for clear goals in mind, and considering the lack of a precise definition of life, in the following section I will concentrate on a particular, and fairly well defined, goal: to create an artificial evolutionary system which exhibits open-ended evolution (as defined in Section 2.5).