For starters resistance changes with temperature.

Also even in a multi-turn potentiometer, getting a precision of 1 in 10^9 would require an equal level of precision in the angle you rotate that potentiometer to (for example, a 0.1 degree error in a 10 turn potentiometer - which I believe is more turns than anything that actually can be bought - translates into a 1 in 36,000 error in resistance, so about 3000 larger than 10^9) even if you had a perfect material whose resistance doesn't change with temperature.

(PS: Just out of curiosity I went and dove down further and to translate a 1/3000 deg movement in a rotating potentiometer into a 1mm movement at the end of a bar attached to it, you would need a 176m long bar - i.e. the radius for 1/(360*3000) of a circumference to be equal to 1mm, is aproximatelly 176 m. This of course has serious mechanical problems even if you remove the bar at the end of the process as the removal process itself would shift the potentiometer by much more than 1/3000 degrees)

The joke here isn't even specifically about resistances and electronics, it's that the real world has all sorts of limitations that when you're doing things wholly in the mathematical world you don't have to account for, and that's a hard realisation for Physicists (having gone to study Physics at uni and then half way in my degree changing to Electronics Engineering I can tell you that's one of the shocks I had to deal with in the transition).

(In a way, it's really a joke about Theoretical Physicists)

See also the "assuming this chicken is a spherical ovoid" kind of joke.