To understand Darwin's work, you have to distinguish between his theory of descent and his theory of natural selection. THe full name of the first is the theory of descent with modification. Some call it the fact of evolution, and some call it the doctrine of evolution.

There is another kind of variation that does not involve the genome at all, and is therefore not heritable. Yet it can produce what looks like evolution. Indeed, the results look so much like evolution that for all we know some of the best examples of evolution may be due to this nonheritable kind of variation. [It] has been observed, is well known, and is well documented.

I am suggesting here that organisms have a built-I capability of adapting to their environment. I am suggesting that to the extent that evolution occurs, it occurs at the level of the organism. This suggestion differs sharply from the thesis of the NDT, which holds that evolution occurs only at the level of the population. Organisms contain within themselves the information that enables them to develop a phenotype adaptive to a variety of environments. The adaptation can occur by a change in the genome through a genetic change triggered by the environment, or it can occur without any genetic change.

Verbal arguments should always be suspect.

Life is so diversified that to any statement I could make about living organisms there are exceptions. Because of the many exceptions, I should qualify everything I say with hedging phrases such as 'generally,' 'usually,' and 'almost always' But I'm afraid the constant repetition of these hedges will slow me down and bore you. So let's make a pact now that I forego the hedging phrases and you are to understand that almost all my statements may have rare exceptions.

Dawkins mentioned two mechanisms: the theory of the ‘primeval soup’ and the Cairns-Smith theory. He discussed the latter in some detail. Since no one has computed, for either theory, the chances of the events occurring, Dawkins could not tell us what those chances are. The mechanisms of both theories, however, have every appearance of being very improbably – even to the point of being impossible.

Information and complexity go hand in hand.