I like how the format forces the papers to be about "big concepts." So many cs papers are about 1% improvements in performance due to esoteric minutiae -- which can't be concisely explained. It biases research towards big picture, heavily-compressible ideas.