But when you record in M/S, you really have one foot in the mono era, or maybe even both feet. The "S" microphone has to be placed with the "M" microphone and pick up its own mix of direct and ambient sound. But again that mixture is critical, and it MUST include a substantial helping of direct sound--otherwise the matrixing function (M + S = L, M - S = R) can't possibly produce consistent directional cues for the direct sound. If your mikes are so far back in the room that the "S" mike is overwhelmed with reverberant sound, then the M/S technique is just churning out fake stereo from ambience information.
By the way, since all X/Y methods (including Blumlein!) are equivalent restatements of M/S and vice versa, the same criteria hold true for them.
--best regards
Mono implies a depth dimension via how dry or reverberant a particular sound is with respect to others within the recording. That depth dimension becomes more clearly discernible in 2-channel stereo, along with the obvious left/right directional differentiation ability which 2-channel stereo provides. The diffuse/pin-point differentiation aspect of 2-channel stereo is just as real and useful as the more obvious left/right differentiation aspects. I consider it a more basic and more important aspect than left-rightness.
Likewise, 2-channel stereo is able to imply an immersive environment beyond the left/right playback stage window between the two speakers, which similarly becomes more clearly discernible and mentally-separable in surround. We can and do work that to our advantage when recording in stereo, and it is arguably more important to do so than for surround (where it becomes easier) just like getting clarity and depth perception correct in a mono recording without everything getting muddied together is more difficult and more perceptually important for good listening results, than it is with stereo.
Even if not ever recording in surround, one can consider these concepts and and apply the ideas in order to make better 2-channel stereo. How can we make it easier for our highly-evolved ear-brains to parse the direct sound elements from the reverberance and audience reaction elements? Record in such a way that the reverberance and audience reaction components of the recording are picked up so as to have greater decorrelation between channels, while the primary musical elements from the PA and stage sources are picked up with stronger correlation between channels. The primary musical elements can then be reproduced with clear left/right imaging stereo cues, while the ambient reverberance and audience are reproduced with more diffuse and cloud-like stereophonic qualities. In that way those different elements sort of remain out of each other's way so as to allow us an increased ability to focus on each part distinctly.
Well decorrelated ambience and audience is not left/right pin-point type of stereo, yet is certainly "real" in a natural hearing sense and not "pseudo stereo". It is a different type of stereo, and the two work together to paint a more convincing picture. Recording in ways such that much of the ambient reverberence and audience reaction is sufficiently decorrelated between channels makes it easier to differentiate sounds arriving from the band in front, reproduced so as to appear within the imaginary window framing the performance within the space between the speakers, from the sounds arriving from all other directions. It's not doing so by actively reproducing the sound outside that stereo window in a discrete way like multichannel surround playback does, yet still conveys it in a way that allows our hearing mechanism to differentiate it more clearly.