Re: A little puzzle
On reflection, it makes more sense to prove the slightly stronger result when you pick N points randomly (with uniform distribution) from the the first M points. i.e. by picking M big enough, you can ensure the probability of the points not being shattered is as small as you wish.
