I've asked Chris at trade shows and discussed it with him at AVS forums, where he had posted for years. My first discussion with him about DSX started over 2 years ago, where he was rather ambiguous about how it worked. Since then he has clarified the process a bit, including the fact that they add reflections.

http://www.avsforum.com/avs-vb/showthread.php?p=16418782#post16418782Hence my point about room treatments and EQ complement each other rather than some sweeping generalization that one is superior to the other. Audyssey knows the size of the area that is being "shown on the screen"? I said they used a different approach. Where did I say it was "better". In fact, I specifically said "Whether it is "better" or not is up to each listener. I don't decide that for other people." We discussed how DSX operates: it doesn't use matrix extraction, it generates reflections that are not in the original recording, it only uses content from the front L/R channels (and no other channels) to generate the reflections, DSX cannot be applied directly to 2-channel sources (they have to be matrixed to surround using PLII or Neo:6 first), height reflections are based on what bounces off the proscenium in concert halls, and other details. As for what DSX can and cannot do: it doesn't know the size of your listening room and it doesn't know the size of the rooms being depicted in the movie, as you have claimed. Without knowing the size of your listening room, it can't make it smaller (it has no reference). And since the processing is based on concert hall acoustics, it can only make your room sound larger (unless you know of any concert halls that are smaller than your listening room). Yes, that's an excellent excuse for making personal attacks.