First because Wolfram is not actually very good at explaining things: he's a very smart guy, but very smart guys don't necessarily know how to lay new concepts out so that other people can grasp them. Second, a MAJOR problem with the book is that he does not admit any sources outside himself: the inventor of CA, John von Neumann, is _never mentioned in the main text_.
You come away understanding only what Wolfram thinks is important, which has a lot more to do with him than with the subject. It's a book worth reading, but it's not an introduction.
(I won't even get into his tortured writing style, for which he actually had to include a little apologia in the introduction.)
ANKOS is sometimes painful to read for Wolfram's incessant self-aggrandising, and it is also frustratingly unstructured in parts. Nonetheless, many of the digressions are thought provoking in ways that you won't really find in any other book. Further, it is mostly non-technical and informal, so it can serve well as a very broad introduction to some of the curious corners of cellular automata research.
This is true. It's a substantial book, well worth reading despite its flaws. The thrust of my post was that it should not be the _first_ thing you read in order to begin learning about CA.