The neat thing about floating point is that it works well to represent real-world values.

It turns out that the real world is amazingly inexact, to the point that the inaccuracies of floating point just never even show up.