Since this discussion, I have been thinking some more about the differences between discretely record Mch sound and simulated Mch derived from stereo or as additional derived channels for Mch. My perspective is from the viewpoint of live concerts in the hall, the acid test of Mch in my opinion. It is not about repanned studio mixes for surround sound, which can be great fun, but which are pretty simple and straightforward uses of Mch.
I think one major difference has to do with exactly what is contained in the signal in the surround channels. In discretely recorded Mch, the more distant hall mikes capture a combination of both direct sound from the stage from that perspective plus hall reflections from that more distant perspective. And, indeed what I hear from my surround channels sounds very much like a more distant replica of the direct sound from the front plus an added ambiance of the hall due to reflections. On playback, this mixed direct + reflected sound field from the surrounds combines with the direct frontal sound in complex ways, modifying the spatial perception via front/back phantom imaging and, to a degree, the tonality. Noticeably, it provides an enhanced frontal soundstage by pulling the image forward into the room somewhat, adding greater apparent depth, as well as widening the apparent soundstage by including side reflections. Meanwhile, the Haas effect keeps the perception of the performers precisely localized up front in space, where they should be.
Note that the surround channels do not just consist of "de correlated" reflections (not sure if I like the term or even know exactly what it means in this context), which add envelopment, although they also, of course, do contain that in discrete. I do not know to what extent various Mch synthesis schemes attempt to recreate both direct+enveloping reflected sounds accurately from the surrounds. I do know that I find them considerably less successful than discrete Mch.
I think synthesis tends to emphasize the envelopment, but it does tend not to provide the added image depth and "pull out" into the room toward the listener, a quality of Mch I greatly admire, myself. The role of the center channel is also important here. A discrete center mike will not exactly capture just a simple sum of L+R mike channels from its perspective, though I suspect many synthesis schemes do a simple L+R center.
I also think synthesis often tends to be somewhat monotonous in its perspective and not adaptable to recorded venues of different sizes and acoustics. Although, there were some early, junky algorithms allowing for manual selection - cathedral, symphony hall, jazz club, etc. - as pioneered by Yamaha, I think, in old AVRs. Some of those are still around in consumer grade HT gear, but generally not in hi end Mch prepros. There are other common, more advanced schemes from Dolby, DTS, etc. My older prepro had seeming dozens of varied synthesis algorithms, most unappealing. I did tend to prefer DTS Neo6 for a time for classical music, as did friends.
Synthesis was fun to play with for awhile, but the novelty has worn off for me. As I said earlier, its appeal for me has been overwhelmed by aggressively increasing the size of my discrete Mch library, now extending to thousands of albums.