So, random error is part of the system (shooting, rifle, ammo, optic, position) dispersion. None random error shifts the entire cone of fire.the target is in MOA I assume because it's close to 1 inch increments which makes sense to people and is easy to measure.
unless I missed your point, my point was human error will possibly double (or more) a math error.
if the math error equates to 5.5" inches at 550yds, (1 Moa for the parallel) a 2 MOA shooter is at 11" with no other error like wind introduced.
Edit: Fumble fingered and hit post before finished typing.
So, a 0.5 mil system at 300 yards will hit a 10 inch target (cone is 5.4 inches at that distance).
You shift the entire point of aim by 0.3 mils and now when you think you are aiming at the center you are actually aiming 3.25 inches closer to the edge, but will still have an ok hit rate as the edge of the 5.4 inch cone only overlaps the edge of the target by 0.9 inches. (17%).
Take that 10 inch target to 500 yards and it is still large enough to contain the entire cone of 9 inches, but start aiming 0.3 mils (5.4 inches) off center and now 4.9 inches (54%) is not over the target.
On the 500 yard example above. If your error is 0.2 mils, you shift your 9 inch cone over 3.6 inches hand 3.1 inches of it is off target (34%).
Your angular error compounds over distance, and thus an acceptable angular error in POI at 300 yards is not an acceptable angular error at 500 yards.
Edit: Percent is calculated off of radius, not area (I'm in the middle of an incline treadmill hike, and not good enough to figure area without paper).