Methodology

Some of the more important methodological issues concerning the scientific use of artificial life techniques were discussed in Section 3.1.3. The key point is that simulations should be based upon explicit theories and assumptions if they are to be of scientific value ([Pattee 88], [Noble 97], [Taylor 98]).

As as example, the emergence of parasites and similar phenomena in Tierra [Ray 91] might appear to be suggestive of a close parallel between this system and biological evolution. However, the appearance of parasites in Tierra turns out to be dependent on some fairly specific aspects of the system's design, rather than on any particular general principles.^3.28 Furthermore, closer analysis reveals that it only requires a single mutation of the original ancestor program to produce a parasite; this was reported in Ray's original paper [Ray 91] (and has recently been reiterated in [Channon & Damper 98]), but has been overlooked in most subsequent reports of parasitism in Tierra. These facts suggest, if nothing else, that claims concerning the relationship between artificial life models and the real world should be treated with caution.

I think that Tierra and many (but certainly not all) of the other recent contributions to artificial life can be regarded as constituting an `exploratory stage' of the subject, in which the potential of the new techniques has been investigated. This is a useful stage, and indeed is a normal aspect of any experimental science [Cohen 95]. However, further progress towards scientific knowledge must be based upon the development and testing of properly formulated theories.