Answer from cs61c-ao (Roger Tseng 16507336) for Question 3
0.5 has more bits so when you do 0.5 - x the bits at the end of the number won't fall off like it would if you did 1.0 - x