The bell curve is a perfectly good working model for a normal distribution but the evil comes when you decide what the variables are and then apply that curve to your population in order to get the result you want.