Why do all compact connected surfaces have a triangulation?

Solution 1:

This paper by Carsten Thomassen (snatched from the MO thread I mentioned in the comments) may contain the most elementary proof.