I found the forum that Larkston Zinaspic mentioned, it's Steve Hoffman's forum. I joined up and attempted to post a response, which will hopefully make it past the moderator(s).
Anyway, here's the guts of what I wrote:
"Unfortunately I was unaware that my Swarm multisub system was being discussed here several years ago, as I would have liked to participate.
"If the topic is still open for discussion, I would like to propose a thought experiment. There will be some simplifying assumptions made, as reality is always more complex than its models, but the idea is to illustrate some of the principles involved.
"Imagine you have a single subwoofer, covering the roughly 1.5 octaves from 60 Hz down to 20 Hz. And let's say that within this passband you have one big (+6 dB) peak and one big (-6 dB) dip at a given listening position.
"Now let's add a second subwoofer to the room, in a significantly different location. Let's assume that it also has one big +6 dB peak and one big -6 dB dip, but because this second sub is in a different location, its peak and dip occur at different frequencies at our listening position.
"The sum of these two dissimilar peak-and-dip patterns will have smaller peaks and dips than either one alone, and more of them (four instead of two). The ONLY WAY that the +/- 6 dB response of the first subwoofer by itself could be preserved would be if the second subwoofer had identical peaks and dips at the exact same frequencies. So with the addition of our second sub in a different location, we now have improved on the +/- 6 dB situation. I think that the most improvement we could ideally hope for in our hypothetical would be to +/- 3 dB, and realistically we'd probably end up somewhere in between.
"Now let's add two more subwoofers. The same smoothing effect would apply, and we'd now have possibly as many as eight smaller-still peaks and dips, though in practice it may not be that many as there would probably be some overlap.
"Anyway the general trend as we add more independent bass sources is a smoothing of the frequency response at any given listening position. In addition, the spatial variation (variation from one location to another within the room) is also reduced.
"I have multiple Swarm customers reporting in-room response of +/- 3 dB across the bass region without EQ, and they are reporting that this improvement holds up over a wide listening area.
"That is the general acoustic theory; there is also a psychoacoustic aspect that we can look at."