I found a lot to disagree with in this book. Mr. Raskin recommends that we dispense with GUI fluff that obscures more than it illuminates (not necessarily a bad idea) and replace it with a system in which your content IS the interface. While typing this review, for instance, I could type the word SAVE, select the word SAVE, and invoke a command to interpret the selected word as a command, thus saving the text to disk. Or type EMAIL (right here in the middle of this sentence!), select the command (and somehow also select the sentence), and tap a key to send the sentence off as an email. Or I can type an arithmetic expression into my text and evaluate it on the fly (which as we all know, most users need to do urgently and often). Truly out-there stuff, and I think that's admirable, but I also think it's wrong. Many of the book's proposed computing paradigms are based on the notion that most files are text files, when in reality, at least in today's systems, only a tiny percentage of files contain human-readable text. We've got applications, MP3s, video, pointers to content, content we've made ourselves, content from other sources. These data are different, and cannot all be tossed into a homogeneous soup and treated as text.
Moreover, the book has some "bugs" which limit its own useability. Mr. Raskin makes dozens of references to a product he designed and extols, the Canon Cat, but never actually explains what it is. I know that it lets users manage files without having to name them (interesting) but I don't know what kind of files they're making, so I can't decide whether this is a good idea. The book does not offer even a single screentshot of this device. Same goes for Swyftware, another oft-cited product with which the author assumes we are familiar, when we are not (Google reports only 7 references on the Web). Instead of showing us pictures these paragons of design in action, the book devotes precious glossy color plates to a gallery of black & white icons, a Windows menu bar, a photo of a grey radio and other illustrations in which color is meaningless.
In this book Mr. Raskin is really thinking, and he does back up his ideas with (talk of) empirical data. And as someone who has developed both hardware and software, he is not afraid to propose alternative input devices and new keys added to the keyboard. That's interesting stuff. But so many ideas just seem wrong. I don't think people want their computer to process keypresses while it is asleep. I don't think people are suffering for lack of a quicker way to enter a Carriage Return character into a search & replace dialog. And I don't think people want to have to learn a command-line interface and then type up their own menus (suffering through syntax errors in the process) to attain the convenience of a GUI. It's a novel book, but I won't recomment it on that basis alone.