John Eargle's book analyzes the localization properties of different stereo microphone techniques. He describes Blumlein, ORTF, and 2 hypercardiods crossed at 120 degrees as the stereo techniques with the most accurate localization.
According to Eargle's book, XY with 90 degree cardioids creates an equally-scaled (across the soundstage) recording that is maybe 20-25% narrower than the actual image; XY with 120 degree hypercardioids creates an equally-scaled image that is roughly 90% as wide as reality. ORTF creates an image that is not scaled and full width (portrays left-to-right accurately); NOS is full width, but pulls the mid elements further to the outside than they actually are. Personally, with ORTF, I am bothered by the sonic effects of the microphones being spaced (comb filtering & slight phasey aspect, if memory serves). I really wanted ORTF to work and even made an ORTF 'ruler' out of aluminum bar stock with 110 degree angles and capsule center marks. But ultimately, it didn't 'ring my bell'.
Taking into account the following negatives, I prefer Blumlein, followed closely by M+S. The caveats with Blumlein are that: 1. it is tricky to get the direct to reflected sound balanced correctly; 2. if the source is wide--like an orchestra--you can't get too close to the ensemble because the back of the Fig. 8's start to pick up too much imformation from the front, which messes with the imaging. To my ear, M+S--when properly implemented (Jensen Transformer M+S box, for example)--does the undistorted left-right perspective thing almost as convincingly as Blumlein and is easier to place because the back of the mic is down in level. much like a cardioid. The caveat with M+S is that there are around 5 ways to get it hooked up incorrectly and it can be tricky to figure out where the problem is. With both Blumlein and well-implemented M+S though, it is uncanny when the assistant claps every second or so and walks from left to right in a semicircle around 10-15 feet out from the mic(s) how your brain 'marrys' the sound you are hearing with the visual image and it's as if the control room glass has disappeared.
Mono compatibility shouldn't be a problem with any correctly-implemented coincident pair as far as I could see; this would not be true of ORTF. Surprisingly, there are still plenty of mono playback devices out there: iPads and theme park rides or parades are just a couple of examples.