You assume that the histogram has some specific layout which it does not. Stops in the histogram are spread out roughly evenly across the horizontal axis of the graph meaning that the histogram is a logarithmic display. That's how you want it to be. Unlike Reichman's implied assertion, one stop does not occupy about 1/5 of the horizontal axis.
Doubling the amount of light does double the output value from the sensor up until the sensor clips.
I never asserted that "expose to the right" didn't result in SNR advantages. I've been practicing it far longer than the LL article has been around and engineers have known this longer than digital cameras have even existed.
Remember that in a 12 bit system, the LSB (least significant bit) represents the lowest stop but that's not the first one. The least significant stop is the last one. Proper exposure sets the most significant stop and the lowest bit is 12 stops down from that. When you look at it in this context you'll understand there is no shadow detail problem unless the exposure is way, way off.
Stops have arbitrary distances on the histogram display and they can vary from image to image. The Kodak SLR actually displays its stop marks, 10 of them anyway, on the histogram directly. They are not evenly spaced across the graph.