Check out this book on Immersive Audio Signal Processing by two of Audyssey's co-founders. Or write to Chris Kyriakakis and ask if he can send the 11 articles he, Sunil and Phil have authored over the last few years on the math and thought processes which led to the Audyssey algorithms.
http://www.amazon.com/gp/product/customer-reviews/0387284532/ref=cm_cr_dp_pt/103-8913234-7281423?ie=UTF8&n=283155&s=books
In one of Chris' papers they describe how, with their very accurate measurement technology they are then able to mathematically synthesize another (I'm guessing here since the text says under 200) 180 points (which makes every 2 degrees) out from around the actual measurement point. Take eight measurements then and you've got ~1440 points to work with.
I'm guessing that, again because of the accuracy of their measurement system, that the multiple measurement points which are perpendicular to the source origin (i.e. a center channel speaker) will coorespond to the + or -30 degree listening window which has been documented by previous audio researchers (Toole, Olive and others before them) as the true width of a human's listening window.
Now, spread 8-10 or 12 measurements (of 180 points each) around a listening area and it starts to become more believeable that a system like Audyssey could indeed optimize a whole area. To my knowledge no other system, Tact included (because I've used it and I know how it operates), works on this principal.