The replicator-first approach assumes that the original seed for life was the existence of some sort of material that could exist in a large or infinite variety of forms, could reproduce more or less faithfully without the assistance of complicated machinery, and had some mechanism whereby specific other reactions or processes could become associated with specific forms of the material. This view was first explicitly expounded by Muller in the 1920s (see [Muller 66]). Candidates for the original self-replicating material of this type include RNA (for references see, for example, [Nuño et al. 95] and [Lazcano 95]), and clay minerals ([Cairns-Smith 71], [Cairns-Smith 85]). The presence of a simple self-reproducer of this nature is enough to begin a process of evolution. The idea is that some forms of the material may be such that they have processes associated with them (e.g. they may act as a catalyst for some reaction) that act to stabilise the material. Such forms will be favoured by natural selection (precisely because they are more stable), and evolution proceeds by selecting reproducers that catalyse more and more reactions that are beneficial to the stability of the replicator. At some point the reactions will effectively give the replicator complete control over the composition of its local environment, at which stage the network of reactions will probably fulfil our criteria for being a living organisation. Many prominent evolutionary biologists and chemists favour this approach, e.g. John Maynard Smith [Maynard Smith & Szathmáry 95], Richard Dawkins [Dawkins 76] and Graham Cairns-Smith [Cairns-Smith 85].
In contrast, the metabolism-first approach assumes that self-maintaining organisations were the seed of life. These models assume that the world is such that self-maintaining (collectively autocatalytic) organisations of chemical reactions occur spontaneously with reasonable probability. Being self-maintaining, they persist for reasonable durations. Another consequence of being self-maintaining is that they produce all of the components from which they are composed, so it is easy to imagine scenarios by which some organisations of this type might reproduce (e.g. by splitting in two). Such self-reproducing organisations will become more abundant, and will replace non-reproducers if there is competition for resources. With self-reproduction comes the possibility of evolution, so any variations of these self-reproducing and self-maintaining organisations that make them more stable will be selected for. By this process, the idea is that a genetic representation emerged by natural selection (in the right conditions)2.17 to give the organisation a high degree of stability. With a genetic representation comes an unlimited range of hereditary variation, enabling the organisations to participate in prolonged evolutionary processes (see [Maynard Smith & Szathmáry 95] pp.67-72). This approach to the origin of life is favoured by Maturana and Varela (see, for example, Chapter 5 of [Varela 79]), and variations have also been suggested by, among others, Freeman Dyson [Dyson 85] and Stuart Kauffman [Kauffman 86].