I know about that. It's bunk. I'll tell you why.
The idea is based around that assumption. That assumption is like the average number of births per woman - 2.3. Yet no one has 2.3 children.
It's an average, and even there it's still wrong, as it's based around a small print (5x7) that isn't even the same aspect ratio of a 35mm film frame.
If you take many people and put them in front of many prints of many types and many sizes, and add in many electronic displays of many sizes, and measure where they stand naturally, you'll get a wide range of distances relative to the length of the diagonal. Maybe that average will be in the 50mm-equivalent range (probably a bit wider), but that still doesn't mean anything - like 2.3 children for determining the best number of children to have.
I like to use the term "natural" for that condition where the final image subtends the same angle of view in your vision as did the original scene to the camera. But there's nothing magical about this condition except, perhaps, in one rare (for us) circumstance - when you should shoot a circular fisheye imagery and project it onto the inside of a dome with your head near the center, so as to create the feeling of immersion in the imagery (those IMAX domes do this, and it's pretty effective). Note that's not 50mm viewed on a 5x7 print.