Real-world 11g overlap
Both Atheros and Broadcom were kind enough to supply spectrum analyzer screen shots that will help show what real 802.11g signals look like.
Figure 4: Real Channel 1, 6, 11 802.11g signals (from Atheros)
Figure 4 shows an composite spectrum analyzer screen shot, supplied by Atheros, of 802.11g signals for the "non-overlapping" channels 1, 6 and 11, with 20MHz channel boundaries marked on the shot for reference. The shapes of the signals look different from those shown in Figure 3 because filtering is applied in the WLAN devices to shape the signal. But you can clearly see that power from each channel's signal overlaps into each other's.
Also notice that each channel's signal is down about 30dB (1/1000) from its peak level by the time it overlaps into the next closest signal. With current radio design technology, this means the overlapping signal has little effect on the primary (i.e. desired) channel.
Figure 5: Real Channel 6 802.11g signal (from Broadcom)
So that I don't get accused of playing favorites, Figure 5 shows a similar spectrum analyzer screen shot supplied by Broadcom. Only channel 6 is shown and the top 20dB of channel 1 and 11's signals are drawn in for reference.
The point of this whole exercise has been to make a simple point:
Key Point #1: Normal "non-overlapping" 802.11b and 11g channel 1, 6, and 11 signals do overlap.
So if normal 802.11b / g signals overlap, i.e. interfere with each other, what's the big deal about Super-G's signal doing the same thing? The issue the amount of overlap.