Focus shift is, as far as I know, an inherent quality in a lens's design. Some manufacturers (Zeiss) will tweak a lens's focus for you to adjust where it focuses best wide open. They did this with the 50/1.5 Sonnar, which exhibits this behavior. The original batch of lenses came from the factory "optimized" for focus at 2.8. That is, the camera's indicated focus at f:2.8 was accurate. At f: 1.5 the lens was front focused slightly, and f:4 rear focused slightly and by 5.6 depth-of-focus covered the shift, so that it wasn't an issue. This was the original compromise around the design upon which Zeiss decided. Well, many folks wanted to use this lens wide open -- that's why they bought an f:1.5 lens -- and Zeiss reportedly adjusted the lens in relation to the mount so that it correctly focused at f:1.5 and then back-focused until f:8. They were not "removing" back-focus as an optical property of the lens, they were just changing where and how it kicked in as the lens was stopped down.
This quality exhibits itself as an unevenness in the zone of focus around the point of intended focus as a lens stops down. So your rangefinder tells you that you are focused on a mark, and at f:4 that particular lens focuses slightly in front of (or behind) the mark. Someone above in the thread said all lenses have it. That is possible, but it is only an issue when the point of best focus shifts beyond the depth-of-focus that your lens has at a particular f:stop and you cannot correct for it mechanically. You can learn your lens, though. I have two of the Sonnars mentioned above and I love 'em both. I just have learned to compensate in use for them. Yes, leaning in or out a little becomes second nature.
If you are shooting a 28mm lens at a subject 10 feet away and do not intend to crop much, it is unlikely that you will notice a lack of critical focus around the eyes, even in a large print. Our own Tom Abrahamsson has used a 28/2 Ultron successfully for years -- you can search RFF for the threads.
Ben Marks
Edit: On Dante Stella's site, he has a review of the original Hexar AF (fixed 35 2.0 lens) -- I think he states that the lens micro adjusted itself after the shutter release was tripped to account for focus-shift inherent in the lens' design at the aperture chosen by the user. Pretty cool trick, if true!