I don't have any experience in plasma manufacturing, but do have experience with firmware that has to compensate for changes in other types of computer systems. My personal opinion would place bets that it is the variations in the plasma manufacturing process causing this. I doubt the firmware is putting out "wrong" values. From my experience, the method used is that engineers estimate the variations, also do test runs on the actual process to come up with the expected range of how something will age over time. Then they select a "safe" value somewhere in the middle to target. Clearly the tighter the manufacturing tolerances, the closer to perfect they can get, but tighter tolerances cost a lot more money! (This may explain why there hasn't been any reported cases of this on the kuro sets, they may have kept manufacturing tolerances much tighter, which they could afford to do with the much higher price tag). Given the low black levels everyone expects, I expect that even very small variations can have a dramatic effect.
So the people that see no or little measured black level changes have a panel that matches the expected change. The people that are seeing the dramatic shift could be caused by several things. Either a) they are at the extreme end of the expected range, b) there was/is some change in the tolerance of panels coming off the line (either a post engineering sample process change or just a plane lapse in quality control) and hence the planned drive voltage change is way to high for these panels, or c) some combination of aging factors that were not considered when the compensation curve was designed.
My own personal opinion is that it is either b or c, and probably more likely b. Either of these precludes an easy firmware fix as some folks are asking for. They can't just arbitrarily change the drive voltage for all panels as it would possibly cause much worse effects (no pixel firings, as you noted in your excellent post). So the notes about having panels with the known problems being sent back to Japan for analysis makes perfect sense, as the engineers would need to see what it was the caused the problem (e.g. where it was off spec), and to see if they can compensate.
Fortunately, unlike some posters that are predicting doom and gloom, I can see that all of these problems have some solutions (compensate for previously unkown factors, tighten manufacturing tolerances, improve QA), although they probably won't come as fast as many would like. Unfortunately problems related to aging systems are notoriously long to debug and implement solutions for. Especially if you want to test that the solution works
Originally Posted by Bdemers
Just to clarify: by "bogus numbers" I mean that I don't personally know what the voltages are. They could be 1 volt, they could be 100 volts. I just made up numbers for illustration. That's what I meant by "bogus."
Now, to answer your other point: I agree to the possibility. It's possible that the variation from panel-to-panel causes this problem, and that to play it safe, Panasonic over-drove the initialization voltage to make sure that no panels had misfiring issues.
However, I've been lead to believe that the plasma manufacturing process is very stable, and that the variability between panels is minimal. We really need someone with more expertise to clarify if the variation between panels is significant enough to cause some to exhibit this issue while others don't.