Replicators don't. Replicate, that is. This is the shocking conclusion to which I have been forced by my attempt to figure out what precisely Richard Dawkins means by the term ''replicator''. Actually, it seems that Dawkins uses the term in at least two fundamentally different ways; but according to Dawkins' own specification of the problem which the ''replicator'' concept was intended to solve (namely, what entities can qualify as things that evolutionary adaptations are ''good for'') then ''replicators'' turn out to be a special form of lineage (what I shall term a similarity lineage); and these, in turn, do not actually ''replicate'' (in Dawkins' sense of the term) at all! Does this matter to the research programme of Artificial Life? Well yes, I believe it does. Dawkins has explicitly argued that there are principled reasons why Darwinian evolution, in any medium whatsoever, must rely on the participation of ''replicators''. Within limits I am inclined to agree. But it follows that, if we wish to realize artificial Darwinism, we had better be clear what a replicator actually is-and all the more so if it turns out that it doesn't...
