What I see "visually", looking into my 13 channel DIY VUmeters, is that in xPropaganda there is more "balance" between the fronts and the sides/top rears. The keyboards and synth in the rear/top does not "hide" the power of the rhythm section in fronts, as it is done with Riverside when the loud guitars appears on sides.
There are many sections in Riverside that the running VUmeters are only the two fronts stereo. The rest of speakers do almost nothing. xPropaganda is more "balanced".
Also the vocals on xPropaganda (at least in the first song that I checked just now) are immersive surround because they output not only in fronts but they are mixed (together) also in side surrounds. That gives an immersive surroundy of the vocals, similar to Bruce in Pineapple Thief. Again, the sounds are more "balanced" and distributed among all channels in xPropaganda than in Riverside. I listen, but I can "see" visually also in the VUMeters.
I do not know if it is just a matter of volume level in Riverside, to "balance" the overall levels better. But I'm afraid it is not just that. There is only stereo, with very small echos in the rest, and when a pure "discrete" sound is coming from sides, like distorted guitar, it is not the same as subtle keywords or synth from xPropaganda. The riverside guitar sound much more aggresive and really "hides" the rhythm section on the fronts. Perhaps it is something like the "size" of the Atmos objects that are very narrow and then appear very discrete in only a speaker, instead of imaging wider in several speakers.
It is supposed that the mixing engineer have to do a listening to check the mix result. Perhaps his criteria, or testing studio equipment, or his ears are different as ours.
I don't have enough knowledge to understand how a good mix could be later spoiled by a mastering with EQ or dynamic compression. I believe, that the mix is poor from the beginning.
EDITED: But, indeed is Riverside and I like a lot the Duda voice. Perhaps I have to enjoy it in stereo