Stephen Jay Gould performs a really unlikely feat in this book; he makes arthropods as fascinating as dinosaurs! In fact he makes a subject that could be extraordinarily dull - the process of taxonomic classification of a bunch of super-old fossils of small, squidgy animals - into a dramatic and gripping read.
He has three major theses:
1.Palaeontology is just as valuable as physics (or any other science).
2.The evolution of self-consciousness was not inevitable.
3.Diversity of anatomy was greatest in the early history of animal life.
As for the first of these - being a physicist, I used to think that physics is "better" than all the other sciences. Then I finished growing up. It therefore came as a bit of a shock to me that Gould complains that physics and other "numerical" sciences hold higher status than "non-numerical" sciences. I figured everybody just grew out of that attitude. Apparently not so. Of course Rutherford said that all science "is either physics or stamp collecting," so maybe I should have realised before now that there are Cinderella Sciences. If anybody should need convincing of how much can be achieved with completely non-numerical science, this is the book to do it. It will also show how difficult it can be - the temperament and set of skills required for the job described here is completely beyond my capabilities.
The second point, that evolution is not a steady march of progress toward "better" anatomies, inevitably leading to big-brained technology-developing self-conscious humans, is convincingly made. Gould considers this a conceptual revolution on the scale of General Relativity, Quantum Mechanics, Genetics, you name it. I was somewhat under-whelmed by this conclusion because I never held the anthropocentric view that humanity is the most important thing in the universe. (Gould never really openly states this, but it's clear that he thinks the trend towards a more secular, less Christian (or monotheistic in general) society is what allowed the breakthroughs described in the book.) I rather think that the conceptual revolution in this case came before the scientific results, not after them.
The third thesis is central to the book and for me the most important one by far. Gould demonstrates that the greatest diversity of basic body-plans of animals was extremely early in the history of animals - right at the start, in fact. Extinction saw that diversity shrink, but the number of species in the remaining basic body plans increase. This pattern is shown to exist on all scales of time and taxon. In other words, according to the fossil record, evolution has progressed by creating more species from fewer basic anatomies that have survived major extinction events.
This confirmed a developing notion of mine that evolution is driven by extinction; the Punctuated Equilibrium Theory that evolution progresses at a much faster rate when something heavily disturbs the ecology could be extended to say "...disturbs the ecology sufficiently to cause mass extinction." I very much doubt that's an original thought on my part.
Just to show that I read this book as critically as Dawkins' deeply inferior The Selfish Gene I will raise a quibble: Gould talks about "successful" taxa. In a "steady improvement" model primates are "successful" because they have big brains, use tools and have complex societies. Gould argues that primates are not successful - there are few species and they seem to be going extinct. Instead, insects are "successful" - over a million species recorded! I would argue that number of species is an equally bad measure of "success" because the concept of "success" itself is inappropriate: One can invent all sorts of measures of success for a species or other taxon - age of the species/taxon, number of individuals, geographic range, etc. Evolutionary "Success" is a concept that should be dropped entirely, in my view.
My purpose in reading books on evolution, is to determine if there is an extant theory that is not obviously wrong and works on all scales of time and space. At one end of the spatial scale there is genetics, but I don't believe that genetic theory is the whole story, any more than the Standard Model is the whole story of physics. At the other end of the spatial scale you have the planetary ecology. On the shortest time-scales we're back to inheritance and genetics. On the longest there is the fossil record as a whole. In between there is mass extinction and diversification of survivors. Linking up all of these elements into a coherent structure and, crucially, explaining how speciation occurs within the timescale suggested by the fossil record would form the theory I'm looking for.