The chemist Alexander Cairns-Smith proposed that inorganic materials, rather than organic, represented the first replicators. The fundamental problem he was trying to address is the requirement that the first "life" (using the term very loosely) had to have been self-replicating. Cairns-Smith speculated that the earliest replicators were not organic at all, but rather were self-replicating crystals that were later superseded by the rise of the far-more-efficient organic replicators.
In this view, the first replicators were crystals of the type that exist in clay or mud along riverbanks; they transmitted their "genetic" information through the natural tendency of these types of molecules to fit together into a geometric pattern. The fundamental characteristic of crystals as replicators must be hereditary variation, or inheritance.
Fortunately, crystals in nature display this pattern: they may be perfectly aligned until a specific point is reached, in which a flaw has accumulated (these are quite common in natural crystals). This flaw has a tendency to percolate down the subsequent layers of crystal, setting up a rudimentary system of heredity. Furthermore, atoms of the crystal's substance may be more attracted to certain geometric patterns than they are to others. This sets up a kind of "differential reproduction" which then leads logically to a form of natural selection.
The hypothetical crystals described above may very well begin a basic process of cumulative selection. Certain crystals may have the property of altering streams or other water sources for their own "benefit", such as by increasing the likelihood of more of the same material being deposited in the same location. Crystals may also encourage the formation of "spores" by breaking easily into subsequent "generations" Those crystals that broke into generations most easily would be selected for; these generations would invariably contain mutations on occasion and would intensify the competition between rival variants.
In time, the crystals could evolve a sort of "phenotype" by altering other materials in their environment. These materials could be used to further the crystal's replication by inhibiting rival crystals from forming or promoting the parent crystal's reproduction. Cairns-Smith's hypothesis is that the materials used by the crystals for self-replication later turned out to be even more efficient replicators in their own right – the earliest peptide-RNA – which ultimately replaced their inorganic substrates.
This process of replacement might repeat for several cycles, or the first products used by the crystals may have been the ancestors of modern replicators - i.e., RNA and eventually DNA.
The principal difficulty with Cairns-Smith’s hypothesis is the fact that clay doesn't necessarily form a lattice/matrix that is perfectly designed for the arrangement of biologically significant molecules. Since there are a rather large number of potential arrangements, getting the precise arrangement necessary to act as a catalyst for a specific molecule is pretty problematic. Finally, the type of clays best suited for this type of “inorganic evolution” are usually found in riparian zones –the smaller biological molecules are fairly unstable when subjected to unshielded UV. It remains to be seen whether such processes could occur in such a way that these molecules could persist long enough to form stable compounds. However, as with Miller, Cairns-Smith’s organic replicator overthrow of the inorganics only needed to occur once…