I designed the Swarm subwoofer system mentioned above, and worked with Jim Romeyn on his conceptually-similar Distributed Bass Array system.
The Swarm was originally designed with dipole speakers in mind, specifically Maggies and Quads. Briefly, dipoles have smoother in-room response in the bass region than monopoles do (according to a peer-reviewed paper by James M. Kates). Smooth bass is fast bass, subjectively and literally - in the bass region, in-room peaks happen where the energy decays more slowly, and vice versa. In fact, the two are so interconnected in the bass region that if you fix the frequency response, you have also fixed the time-domain response. So the key to matching the "speed" of dipoles is, match their in-room smoothness.
Two subs intelligently positioned will be roughly twice as smooth as one sub. Still not as smooth as two dipole main speakers, but definitely better than one sub. Four intelligently positioned subs will be twice as smooth as two, and comparable in smoothness to a pair of dipoles.
The improved in-room smoothness that arises from using multiple subs intelligently distributed may seem counter-intuitive, so let me try to explain: Imagine you are looking at the in-room frequency response of a single sub. You see a few big peaks and dips. The problem is not only their magnitude, but also how far apart they are - you see, the ear/brain system tends to "average out" peaks and dips that are within about 1/3 octave of one another, but these are almost always going to be further apart than that. Small rooms are typically worse than big rooms in this respect.
(Also, the ear is especially sensitive to changes in loudness - peaks and dips! - in the bass region. This is shown by the way that equal-loudness curves bunch up south of 100 Hz. So there is usually a LOT of room for improvement in the bass region.)
You can move the sub or move the microphone location and those big peaks and dips get shuffled and changed, but they do not go away. Now imagine that you add a second subwoofer in a different location. It too generates a big peak-and-dip pattern. But unless the second sub’s peaks and dips overlay the first’s (which can only happen if the two subs are in the exact same spot), the sum of the two will be significantly smoother than either one alone. Mathematically, it will tend to sum to half as much average deviation (or "twice as smooth"). The same trend holds as we add more subs. And not only are the peaks and dips smaller in the summed response, but they are also more numerous and therefore closer together, so the ear/brain system’s averaging characteristic comes into play, and the subjective improvement is probably greater than one might think from eyeballing the summed curve.
Anyway my point is, there is some solid acoustic and psychoacoustic science behind the distributed bass array concept. Credit to Earl Geddes for sharing his ideas with me and allowing me to use them.
Duke