While at CVPR 2006, I noticed that a lot of people wanted to tackle big Vision problems – things like getting a whole scene out of a single image, or constructing object finders out of sketches. Marr had 18 citations, so I finally got around to reading the whole thing – not just skipping over the parts that seemed out of date or just plain wrong. And a lot of it is wrong, with other books like Forsyth and hoary Horn providing a better technical idea of what how the field is shaping and how to put the basic ideas to use. However, Marr is a good introduction to ideas about how research in vision, and other such fields, should be done, if not in any usable form today. Next, I’ll read Aliomonos’ old thing. If I was in Biology, I get the feeling this would be reading Darwin after reading Gould, or whoever the hot one is now.