... based on the first few chapters, without much commentary for now (I'm still reading).
0. Darwinism is defined as common descent + random mutation + natural selection.
1. Behe believes in an old earth.
2. Behe believes in common descent (on the basis of compelling scientific evidence), and natural selection (on the basis of it being trivially true).
3. However, he doesn't believe that random mutation is a sufficiently powerful mechanism to produce either complex structures, or the apparatus that engineers those complex structures in the cell. So in terms of where the "intelligent design" happens, he would presumably suggest it was in the presence within cells of DNA that encodes these complex structures - whether this is "front loaded", or inserted at some later stage in the development of life.
4. He spends time defining the sort of change that darwinism could achieve. The examples he gives are thousands of years of battles between the malaria organism and humans. Malaria has managed, through single or double changes to bases in DNA, to defeat the attempts of human intelligence to artificially wipe it out. And yet Malaria has not managed to find a way of overcoming the sickle cell mutation that conveys protection for humans to the organism. This is again a mutation of one base. There is a mutation of a second base that conveys the same protection against malaria, without the harmful effects of sickle cell disease. And yet this has not become widespread amongst humans. He spends some time talking about what this means in terms of population genetics, in "layman's terms".
He also talks about the fact that the evolutionary explanation for the appearance of antifreeze glycoproteins in notothenioid fish is reasonable. As, for that matter, have I - and I feel quite smug that I should have identified the means of fish reproduction as assisting the evolutionary process in this regard, though nobody else has picked this up yet! A series of small, relatively high-probability evolutionary steps allows this to occur. And yet, the malaria organism, which would also benefit as a species from being able to move to cooler climates, hasn't done so, despite a population of 1018 organisms in the world at any one time (around a million infected humans, around a trillion organisms per person infected). And in any case, the evolutionary change to notothenioid fish could be described as an "additive" - no complex new structure has appeared.
5. In terms of the bacterial flagellum, he asserts that, whilst attempts have been made to provide a narrative explanation of how this might have evolved, none of these are really satisfactory, and they fail particularly to explain the complexity of the mechanisms that control assembly. He also points out that the cilium is probably an order of magnitude more complex, and talks about the engineering detail in its structure.
6. The aim of his book, then, is to define "the edge of evolution" - the sort of changes that evolution would be able to achieve through CD/RM/NS, and the sort of changes that evolution can't achieve, as the steps are too great. Before anybody says in an anthropic way, "No steps are too great, because we are here," Behe has made the point that there are some evolutionary changes, like malaria overcoming the sickle cell mutation or its temperature limitations, that are just the sort of things one would expect to happen in evolutionary terms, and the malarial organism population is large and able to try lots of variations to great effect.
The issue of the edge of evolution is also one that I have explored before - do a search for "specification" in this blog. Basically, I was saying that if too many low-probability events are required for evolution to proceed, then it won't proceed. I guess that's where Behe is going - I'll let you know.