Computers are getting a hard time removing the error in a simple computation, but have a hard time trying to represent imperfect things from real life...
Update: Seems to me, no one has this wright (and probably never will because of that floating point binary representation)