Most people who write books about computation starting at the level of assembly language would then work up from there; Feynman works down, covering lots of the fascinating nitty-gritty stuff that he, as a physicist, is interested in. So there is lots of material here that is rarely found in computing texts, and certainly even more rarely found in as accessible a form as this. And the lectures that are right down in the physics -- on thermodynamically reversible computation and quantum computing -- are some of today's hot topics: Feynman, as usual, was way ahead of his time.
This is a write-up of a series of lectures Feynman gave at CalTech in the mid 1980s, transcribed from tape recordings. So the chapters capture the flavour of the great man's lecturing style, and the informality of the spoken word. But although I am a great admirer of Feynman's, I don't think the change of medium works too well in this case. I'd love to hear these lectures, but when reading, I would prefer a deeper and more polished form.