If I'm too close to be getting good sound from that second stack (greater then 35 degrees in my experience), then I give up on the second stack and point both trending toward the one good stack.
I think that's very good advice, especially using standard configs.
Let me try and better explain my thinking behind the seemingly off the wall suggestion for parallel figure 8's- The reason I suggested such an unusual configuration is exactly the insight Page notes above. There are two unusual things about this config which could be advantageous in this situation:
1) Rejection. Because both mics are oriented parallel to each other, without any angle between them, their polar patterns are aligned. That is important because their composite stereo polar pattern, the combined pattern of both mics together, becomes essentially the same as one mic alone. That's very unusual and doesn't happen with any standard configuration except spaced omnis, in which case the individual mic orientation doesn't really matter. What makes that especially useful here is that the region of minimal sensitivity (the figure-8 'null', 90 degrees off-axis) of both mics line up with each other, and is located as far 'forward' towards the front of the polar pattern as possible. That means as a combined stereo pattern, this configuration has the most off-axis rejection to the sides of any other possible stereo configuration by far.
2) Left/Right stereo balance between channels. Any stereo mic'ing configuration that uses an angle between directional mics exaggerates the level difference between channels when recording from an off-center location, regardless of which direction the stand is pointed. The larger the angle between mics, and the farther off-center, the bigger the problem. With the polar patterns parallel aligned, signal level decreases evenly in both channels as sound arrives from off-axis. That helps significantly by simply not making the problem worse than it already is. A far less significant benefit is that by pointing the array toward the far side, the close, ‘too loud’ stack is then more off-axis to the mic pair and it’s direct sound is picked up with slightly reduced sensitivity. That effect is probably insignificant with any pattern other than a figure-8 arranged at a pretty large off-axis angle to the near-side stack. For a decrease in level of 10dB for the near-side stack (giving it half the perceived loudness it would have if on-axis, to better match the level of the far side stack that is on-axis) the near-side stack would probably need to be about 70-80 degrees off-axis. That would not be typical in many situations, but does apply to the unique situation described in my first post.
It's those two unique attributes which makes this parallel-8 configuration an exception (possibly the only one?) to "Page's guiding rule of stack proximity".
I think the reason why POS can work better than other typical configurations when far off center and /or further back is that it typically uses less angle between mics than other ‘standard’ configurations, which helps by reducing the near-side louder problem and reducing the overall stereo pattern sensitivity to off-axis room sound. In that way POS is a move towards a spaced, parallel configuration. What would be beneficial for most POS setups is making up for the reduced mic angle by introducing more distance between mics when possible. I think people recording from far back (off-center or not) in situations with compromised acoustics might do well to experiment with parallel spaced directional mics.
Hypers PAS
^^
..with a bit more space between mics than is typical.