let me put this in a different way, in a way that maurile often puts it better than me.
by reverting to the mean, it means that he is reverting to the expected average production for HIM regardless of what other QBs do. now it is probably prudent to use other QBs in a historical context to set up parameters for your projections, but ignoring that let's take the compiled projections as his % chance to produce at these levels.
for example for passing yards, we expect him to have a:
1.84% chance to pass for less than 1,000
2.45% chance to pass for 1,000-1,499
7.98% chance to pass for 1,500-1,999
30.67% chance to pass for 2,000-2,499
42.94% chance to pass for 2,500-2,999
11.04% chance to pass for 3,000-3,499
3.07% chance to pass for 3,500+
over the long run, assuming for the sake of this example this is an accurate assumption of what his typical year would look like, he will in most seasons be in the 2,000-2,999 yard range. there is a slight chance of not producing at all (maybe due to injury) and a slight chance of having a monster year.
if he has a monster year in 2007, it is likely he will revert to his mean as him duplicating this 3.07% chance two years in a row is extremely difficult to accomplish.