2N Noncollinear Points Determine at Least 2N Directions, by Peter Ungar. This is a beautiful short paper that proves the result in the title.
A general remark: If you have to chose choose a single paper (or a single paper of a mathematician selected in other answers), I would recommend more strongy to choose original papers of an important basic results rather than a large survey paper papers or a "meta" paper about mathematics. (This is also closer to the original intention of the question.)

