Natural selection works on populations of organisms, not on a first self-replicatiing organism. it would only be when that first self replicator has produced a population that natural selection can work. Intervention may not be needed. Twenty years ago Thomas Cech discovered sequences of RNA that could self-catalyse and a similar discovery was made in Australia by Gerlach et al soon after. These sequences were less than 15 nucleotides long. While there is no particular reason to say so, it may be that a sequence of 25 nuclotides may be able to self reproduce. If there are four nucleotides available for every position in the sequence, the number of possible sequences is about 1 x 10^15. A handful of mixed ribonucleotide monomers weighing 120 grams has about 6 x 10^23 monomers. If these are polymerised at random to a length of 25 monomers, then, in the absence of any chemical bias, something like 6 x 10^8 examples of each sequence will form. If you extend the needed length to 39 nucleotides you still get 2 examples of each possible sequence. Now consider the fact that there are something like many millions of stars in this galaxy alone. Recent discoveries have revealed more than 300 planets orbiting nearby stars. Only one of these is anything vaguely like the Earth, but it is an indication that planets are not rare and therefore there may be millions of planets something like the Earth in this galaxy alone. Multiply that by maybe a billion for the the other galaxies. Now run the sort of chemistry I have roughly descibed on each of these planets throughout the entire Universe every second of every day for a million years. To me it's beginning to look as if life is not rare, it is inevitable given suitable conditions.

The very origin of life has nothing to do with evolution. It's called abiogenesis and Palifox has described it very well.

To put it another way, natural selection is only one mechanism of evolution.