Supposing we have 2 circles in a 2 d space that can bounce off each other like balls. These circles are made of an infinite number of points. We put the centre of each circle on a line and send them towards each other along the line. Are they going to bounce off keeping their trajectory on the same line? The derivative on the point of impact is zero and I expect they will return on the same line. However, if we want to place each circle on the same line, we will need to measure if it is in a right position with a certain accuracy. Say one circle starts from the left and the other from the right. If the accuracy is a finite number, I suppose the left hand circle will always bounce in 50% of the cases up and 50 %down. How can we tell the difference between precision of measurement and the world intrinsic randomness, as QM would suggest?