One would be wrong. This book goes into rather impressive depth on some rather abstract concepts of computer science without dabbling for too long in the details. It does the best job I've ever seen of explaining the Turing machine and how it relates to computability and decidablity.
The exercises are both easy and insanely difficult - so you can basically chose your level and then go through the book, some of the problems are very hard, some are trivially easy, a great mix makes for great homework assignments.
The "Proof Idea:" sections before every proof give you the underlying concepts in plain english that are about to be stated formally so you have a clue what's happening when the formal definitions start flying. These are priceless and should be included in every other book that uses formal proof techniques.
The book reads fairly well on its own, or makes for a great class text book, which I used it for. As my professor said, "This is a good book because it doesn't have any extra words." but you don't seem to mind as you read it. Probably the best work on the science of computation in the world, certainly the best I've ever seen.
Proofs on theorems are given virtually always in two steps: first you're presented with the idea that lies behind the proof, and then you get the proof itself in a more rigorous fashion. Again, Sipser strikes here because it's harder NOT to understand one of his proofs than the contrary simply because the presentation is always clear and understandable.As a matter of fact, Sipser (as he point out in the preface) almost always avoid to overload proofs given by construction with more rigorous following proofs (e.g. induction on the constructed machine to prove its equivalence with ...). This has a strong impact on the attention you can keep when studying throghout a chapter: avoiding to dive into tedious details when the proof (by construction) has been clear enough help to keep you attention high and boredom away. This is a way of learning, an effective way.
Sipser uses sometime a notation that's different from the somewhat standard one (e.g. the description of delta or transition function on various machines), but it is coherent throughout the whole book, and that's what does count, together with the note that this notation is noway more complex or hard to understand than the "standard" one.
Should I name two books on Theory of Computation (not Complexity), one just a little less rigorous and one just a little more rigorous than this, I would suggest Coehn's "Introduction to computer Theory" and Kozen's "Automata and Computability" respectively.
My conclusion is that this is a great book, worth the price (especially if confronted with others ...) and a stable place in my bookshelf.
As the two courses progressed I began to wonder why the graduate text was so much better than the undergraduate text. Later, I checked out an Amazon review for the undergraduate text and discovered that the reviewer was as disgusted with it as I was with the book selected for the undergraduate course. Further, the reviewer specifically recommended this book.
This text really is geared for the advanced undergraduate or graduate student. However, it is beautifully thought out and written. Consequently, it is a far better undergraduate introductory text than several of the books which pretend to service this market.
The author really REALLY did a great job in clear, simple explenations.
What I liked most is the way the author presents his view on how problems should be approached, when to use this strategy and the useful examples - which really are explained step by step.I didn't see anything "left to the reader" - so the examples really help out in understanding the fine details.
Sipser uses clear, short proofs to theorems - which makes them easy to understand and to remember, and I might even say intuitive.
Never have I studied a subject so fast, and so thoroughly as I did using Sipser's book.
