Device Under Test
I think that all 3 approaches discussed in the diyaudio thread are valid. The end result is the important part. Reducing the seat to seat variance WITH as little EQ use as possible is the goal.
With the Welti approach (symmetrically placed subs), it is done using standing wave theory that assumes a rectangular room with essentially identical walls. The drawback is that rooms rarely fit this category, but the approach works quite well in many instances. The drawback is that although the response is nearly identical at each seat, lots of EQ must be used for a flat response, and the overall LF contribution of multiple subs is not always realized, especially with 4 subs at wall midpoints.
The Geddes approach is more experimental, and also provides for cancellation of some floor/ceiling modes. By experimentation and measurement, a solution can be had with less subs.
The Parham approach is between the two, with an overall increase in LF radiators.
All three approaches are valid. The endpoint is of course measurement. Although none of the approaches breaches the subject at length, absorption on front and rear walls can significantly affect 1/4 wave suckouts in the 100-300Hz region. To a certain extent, absorption can also help with <100Hz content, but not nearly as much as mode cancellation with multiple LF sources. A combination of the two + DSP control (absorption + multiple sources + DSP control) can yield a very flat freq response and little seat to seat variation, as Toole experienced in his own living room (Room 3 at the end of the paper):http://www.aes.org/tmpFiles/elib/20111227/13680.pdf
Start at section 6 of the paper, and this is the basic algorithm used by the JBL BassQ unit. It allows for optimized use of delays, levels and a single band of EQ per sub to reach a solution that allows for the least variance seat to seat plus as flat a freq response as possible using minimum EQ. A disadvantage to the BassQ system is that it only allows for four measurement locations (you use four mics simultaneously during setup). It also does nothing below 20Hz, a problem for the ULF-loving people here.
Since there is no established algorithm on how to do a BassQ type optimization with a laptop and single mic, experimentation and measurement are key.