I'm pretty sure that the challenge is to prove that you can for any coloring construct a finitely defined rule for picking the members of the subset which is guaranteed to give you a monochromatic subset.
The question is more about if you can always find such a (finite) rule to partition the set, rather than if you can in a few easily constructed examples.