This is one manifestation of quantum uncertainty on the macro scale - one that scientists have failed to adequately study. You see, there is not a fixed number of screws/nuts/etc in a pinball machine. The possible values for the number of screws/nuts/etc in a given PF/assembly/etc can only be represented as a probability distribution. In fact, (in a manifestation of the Heisenberg uncertainty principle), you cannot measure the number of parts in a given example of a pinball machine (or subassembly) without taking it apart and counting them. And when you put the machine back together, you similarly cannot know exactly how many screws/nuts/etc will be required until you try to put it back together - though in all cases the number of screws/nuts/etc will be close to what was specified in the design.
All this assumes that any example of a PF or other subassembly is working - if not, of course, all bets are off w/regard to how many parts are in it.
...
Yes, I am a bit bored at work.