The researh group I'm in covers human-computer interaction, as well as graphics and hardware, and this seemed a relatively popular book. On my recent holiday in Egypt I found it in a book shop, and decided to get it as I felt I was a little low on reading material.
It calls itself "New Directions for Designing Interactive Systems", and it's certainly not about working to the current standards. In short, it looks at a move away from the current paradigms, towards more elegant, modeless designs. The shift is large, and I think some of the work is worth taking with a pinch of salt, but it shows what interesting ideas can be produced if you do not constrain yourself to current systems. (Freakily, the UI I've used that most resembles Raskin's ideal is... Emacs!)
The book is useful as inspiration for new interfaces, but it is not useless if you have to work with existing systems - it encourages critical thinking about interface design, and some of the ideas from the book may be applied incrementally to existing systems. The downside is that it highly focused on document-creation (and to a lesser extent, viewing), which does seem to leave a blindspot in some other areas of interaction.