I was leary when I first started, on account of the (for me) immensely technical nature of the first couple of chapters, but I found the book as a whole very interesting.
The book's primary shortcoming is, I think, unavoidable. It certainly succeeds at bringing the material realities of electronic texts into the larger textual studies conversation, but that achievement comes with a great deal of internal tension. Kirschenbaum's subject matter requires him to discuss a lot of subject matter (the technicalities of magnetic encoding technologies, usage of a hex editor, the binary coding behind the ASCII character set, etc.). His arguments have to assume a certain level of technological knowledge. Yet, his book is directed primarily at an audience of English professors. He makes some valiant attempts at explaining things without becoming a primer on programming, but, as a liberal arts person, I still found myself confused a lot of the time. I understand the basic principal that all things electronic also have a physical dimension, but the finer points of the discussion were lost on me. Periodically, there would be a screen shot of, for example, Mystery House opened in a hex editor. Even toward the end of the book, the only response I found myself able to have was something along the lines of “Hmmm. A long string of unintelligible symbols.” As a reader, it's frustrating, but I have to admit that even alerting me to the fact that there is such a thing as a hex editor is no small acheivement on Kirschenbaum's part. Not that I understand what, exactly, it is that a hex editor does. But at least I know that they exist and are vaguely connected with underlying code.
Overall, Mechanisms has its frustrations (besides the technical terms, I'm convinced it could have lost 50-60 pages without suffering much ill effect), but it also has plenty of interesting things to say.