I found this book an excellent deep and thorough analysis of some of the questions regarding evolution theory. I would highly recommend this book for those interested in the topic.
Did Darwin write the origin backwards:
Sober carefully develops how natural selection and common ancestry are distinctly different ideas and that natural selection is causal but common ancestry is the most insightful concept in Darwin's book "Origin of Species". So even though common ancestry is more important, Darwin started with the causal argument. The concept he develops of natural selection being the force causing the branching of the tree of life is very good and important. He also makes the important point that natural selection is not the only cause for inherited traits but is a very important and the fact that it is not the only determinant does not weaken the argument for evolution as described by Darwin. The discussion on Darwin's principle that adaptive traits provide almost no information about ancestry whereas as non-adaptive traits provide strong evidence was well made and an important concept. He demonstrates that this principle is not always true but is true more often than not. He builds his arguments on very careful probabilistic logic in a very clear way. The only thing I did not like about this chapter is his going along with the idea that there was probably only one ultimate common ancestor following the typical arguments going against Darwin's more lenient perspective, e.g., the fact that there is mostly one genetic code (not true in every case) is a strong argument. There are a number of these arguments that support the one common ancestor idea, however, they are all based on the current living descendants. Are we to believe that somehow an ancient ancestor with a full genetic code suddenly appears and this is how life originated? I find that very unlikely. What is more likely is that multiple entities capable of metabolizing, growing, replicating with error may have preceded an ancestor with the genetic code leaving no evidence that we have been able to discover so far. There is also the possibility of multiple entities with varying genetic codes that were reduced to a surviving ancestor who we are now descended from similar to the mitochondrial Eve hypothesis for human beings. This is the cautious side of Sober and is completely understandable although I like Darwin's bolder approach on this.
Darwin and group selection:
This is an excellent chapter on not only group selection but also multilevel selection. He begins by taking apart the arguments of George C. Williams who started the counter argument to group selection in his 1966 book. One by one, Sober provides a very strong case for each of Williams' five major arguments. He extends this argument to the conventionalists who have also used kin selection and inclusive fitness (in the postscript) using very strong and carefully laid out probabilistic arguments against their denial of group selection. He also takes apart Dawkin's selfish-gene'ism and adaptationism that essentially bought Williams arguments totally and made them popular. He carefully analyzes Darwin's arguments on human morality, social insects with infertile females, and the infertility of hybrids, not all of which Darwin got right. One disappointment is that this section is dated not including concepts like epigenesis, developmental plasticity, and genetic accommodation which reintroduce the Lamarkian like possibility of inheritance of adaptations of the adult to environment. I have been convinced by his arguments and somewhat embarrassed by having been influenced in the past by very incomplete and even non-scientific arguments against multilevel selection/adaptation.
Sex ratio theory- Darwin before and after:
Sober starts by taking apart Arbuthnot's faulty argument that the fact (turns out it was not actually fact) that there was a male bias for births in London resulting in equal numbers of males and females at reproductive age was proof that God was in control since monogamy is moral and polygamy is sinful. He goes from there to the evolution of this argument through several stages of analysis historically (Dusing, Fisher, Hamilton). He continues to build the case for multilevel selection which he completes in his postscript.
Darwin and Naturalism or why science should stick to methodological naturalism:
Sober defines nature and supernature as "the totality of entities, events, and processes that have spatiotemporal location; supernatural entities, as I'll use the term, do not." In the Notes section he adds: "I'll try to avoid saying that supernatural entities are "outside" of nature, as this suggests that nature is a box and that supernatural things have a spatial location-they are outside or above the box."
This is not the common definition that you would find and avoids defining supernatural processes. Some definitions from Wikipedia for supernatural include:
"is that which is not subject to the laws of physics, or more figuratively, that which is said to exist above and beyond nature."
Under the subtitle Catholicism in the article on supernatural: "the "Supernatural Order" is the gratuitous production, by God, of the ensemble of miracles for the elevation of man to a state of grace, including the hypostatic union (Incarnation), the beatific vision, and the ministry of angels." And "The supernatural order is then more than a miraculous way of producing natural effects, or a notion of relative superiority within the created world, or the necessary concurrence of God in the universe; it is an effect or series of effects substantially and absolutely above all nature and, as such, calls for an exceptional intervention and gratuitous bestowal of God and rises in a manner to the Divine order, the only one that transcends the whole created world..."
Also "The term "supernatural" is often used interchangeably with paranormal or preternatural -- the latter typically limited to an adjective for describing abilities which appear to exceed the bounds of possibility."
The supernatural commonly includes the claim of an immortal soul for all people, the existence of angles, Jesus as the son of God, etc. These appear to have spatiotemporal location but most generally violate natural law, i.e., what natural science would predict to be possible and in very obvious ways.
Sober's methodological naturalism argument that science has no basis for evaluating the supernatural is based on his very weak amorphous definition. Instead of the supernatural as something that might influence evolution on the margins of determinism such that it could not be detected by science, it is generally proposed to be capable of outlandish accomplishments that are not possible. In those cases, science can obviously be utilized to test the question of whether something was supernatural or natural since the event by definition violates natural law, the domain for science. Even in the case illustrating the idea of supernatural influence on coin flips, it is quite possible to determine if a supernatural being is influencing the outcome of multiple flips of the coin unless the supernatural entity is adding random effects. What possible purpose would any supernatural being have for adding random changes to the initial conditions of a coin flip? Generally, religions pose supernatural beings shape the processes of nature to achieve their purpose and mostly by very dramatic acts. If we are to conclude that science should stay away from questions regarding supernatural beings that either are totally uninvolved (deism) or that are so subtly involved as to be undetectable by science, than what is the point of even hypothesizing such a being? What possible meaning would such a being add to life? Essentially supernatural beings that are this wimpy are pointless and it would be a waste of a scientist's time to spend effort disproving their existence.
On the other side of the argument for methodological naturalism, Sober argues that religious argument should stay out of science, especially the creationists arguments. I like this argument since this is a methodological divide, i.e., religious arguments are not scientific arguments.
He has written three: "Second thoughts about cladistics parsimony and the tests of adaptive hypotheses", "More on units of selection", and "Evolutionary theory and the reality of macroprobabilities". It is not clear why this specific collection other than the section on selection supports especially his arguments for multilevel selection/adaptation. Still, I learned from this some the deep assumptions made in evolution arguments that are often not explicitly stated but have definite implications for interpretation. The last section was especially dense for me concluding "the reality of microprobabilities entails reality of diachronic macroprobabilities." Which I interpreted to mean that reductionism is not inherently more objective or true than holistic analysis.