Distance from the front not the rear wall?

The simple explanation is that

1) Bass response is always the worst close to a wall - just try it by walking around the room - you always get a more uneven bass response when you sit near a wall.
2) The secondary reflections will collapse the soundstage to the speakers - just try it by moving your position out into the room and you will find the soundstage grows in height and width and frees itself from the speakers. However, with the back of your head against a wall the sound collapses to the speakers.

Depending on your setup this will be more or less apparent. You need to have the speakers free from nearby reflections to begin with to hear 2.

The problem with Sonics explanation above is this statement:-

"Secondly, the reflections are shorter than the circumference of the head, so the brain cannot measure the time delay between the ears, and therefore cannot localize the source of sound. When the brain cannot localize reflections it ignores them."

This is obviously not true for a listener seated at 1 to 3 feet from the wall behind them, as suggested in the article). Physically the reflected sound has to travel at least two feet further than the direct sound when the listener is 1 foot from the wall. As the article mentions at the beginning, this is very bad because of the 5 msec rule which requires that NO reflections reach the ear prior to 5 msec.

If you apply the 5 Msec rule(which is scientifically proven) then you need a minimum 4 feet from the speakers to the side walls and at least 4 feet distance between you and the rear wall behind your head (as I recommended).

Why do I say 4 feet and not 5 feet (since 5 feet is about 5 msec)?

This is because the diagonal path that reflected sound follows will almost always be longer than the physical distance because the sound has to "reflect". You can, in some situations, get away with 3 feet.

The inconsistencies in the article by Joachim Gerhard should be self evident - although a lot of what he says about speaker placement is valid. Although, such a wide speaker separation tends to exaggerate the stereo effect often to the detriment of a more natural presentation in favor of a more impressive presentation.

IME, with speakers out from nearby walls (especially sidewalls) and absorbing room treatment behind the close wall to the listener, I enjoy better sound than from other arrangements.

One has to work with the room he/she has. If it allows for enough space behind the speakers, behind the listener and to the sidewalls, that is great. If it does not, one needs to adjust. Jim Smith just voiced my system in my small to medium sized room. The first thing he did was to locate the listening seat for best and smoothest bass response. In my case, this meant moving the sofa even closer to the rear wall. My ears are now about 12" from the wall behind my head, but that wall is treated with absorption. Bass response, both in extension and smoothness, as well as soundstaging and image focus all improved with deliberate and careful placement of the speakers, the listening seat and specific room treatment.

I think it is very hard to generalize about this stuff. One does want to minimize reflections and give most speaker designs space to breath, but beyond that, each room is different and each system is different, and it will take time and experimentation with placement to optimize the sound in a given room for the most realistic presentation.

I think it is very hard to generalize about this stuff.

If everyone accepts that speakers are best kept away from reflective surfaces by several feet (ideally 3 or 4 feet or more in order to get an optimal stereo image and soundstage) then according to physics it is a given that the same applies to the listener. (A reflected sound within 5 msec of that of the direct signal from speaker to listener can be created either way: Speaker=>near speaker side Wall=>listener or Speaker=>near listener rear wall=> listener; and in both cases the imaging will not be as precise as without these close reflections)