I believe that societies — collections of socially constructed ideas — evolve, and they evolve by virtue of a dialectical relationship between heritable variation and natural or artificial selection. However, this view does not entail or endorse the 19th and early- to mid-20th century idea of Social Darwinism and ; indeed social evolution is almost completely different from the latter two ideas.
First, Social Darwinism is fundamentally a moral position: that "inferior" human beings "ought" in some sense to be killed or otherwise prevented from reproducing. The sort of social evolution I'm speculating about is a scientific theory, and scientific theories do not in any way entail fundamental moral positions. A scientific theory can tell us how to achieve our choices, but it cannot tell us what we "ought" to choose.
Secondly, both Social Darwinism and Eugenics are genetic: they attempt to select against "bad" genes. Social evolution, on the other hand, describes change in the kinds and frequency of ideas.
Besides being prima facie morally objectionable, effecting social change by modifying human genetics by artificial selection is impossibly slow; it's easy to see empirically that our ideas have changed just over the last 10,000 years or so of recorded history far faster than even the fastest, most efficient artificial genetic selection would permit. Additionally, we have seen from our experience with animals that artificial selection can efficiently produce only single-purpose sub-species. Human beings, however, are general-purpose beings; a single-purpose human being is useless: we can usually achieve the same effect orders of magnitude more efficiently by creating new machines. new technologies or even entirely new technologies. It was, for example, much more efficient to create — from a standing start 10,000 years ago — logic, philosophy, the scientific method, classical physics, mathematics, quantum mechanics, semiconductors, and computers than it would have been to breed human beings who could perform arithmetic as quickly and reliably as a pocket calculator. Effecting social change by artificial genetic selection is like trying to repair a pocket watch with a sledge-hammer.
The specific physical mechanisms of variation, heredity and selection in the pool of ideas is almost certainly very complicated, and deserves careful, detailed and expensive scientific investigation... investigation that I personally am utterly unable to perform. But just a cursory examination of history available to any educated person is enough to reveal some obvious truths.
Most importantly, we can see that actually killing people (notwithstanding our moral objections) is a completely ineffective way to directly select against the ideas that people have. To gain even a small selective effect, you have to kill enormous numbers of people very quickly, and you have to make sure you kill all the people who have the idea. I think the last time killing people to effect social selection actually worked was in the massacre of the Albigensian. Neither the Inquisition nor the Holocaust even made a dent in Judaism.
Not only that but, human beings being who we are, killing a lot of people seems itself to select against the ideas of the people doing the killing. There aren't very many Nazis now not because we killed them all in 1945, but because most people reject (i.e. select against) the brutality, bigotry and obvious stupidity of their ideas.
So please keep in mind that when I'm talking about social evolution, I'm not talking about anything resembling Social Darwinism or Eugenics.