>>2
You can create a well-ordering of girls judged by their looks if you compare every girl to some other girl just once, which is
n–1 where
n is the number of girls on campus. It is impractical because if you split the work between a number of assistants, local cycles in your ordering are likely to be created (notion of transitivity of the ``more ugly'' relation is sometimes lost between males, and even more so between females), which will force you to take a statistical approach which we will describe in our next paper.
If you cover only a small part of the
n2 comparisons (of every girl to every other girl), you still get pretty accurate results, and the probability that your current ordering is the natural well-ordering approaches
1 asymptotically.
Please note that doing so and publishing results of your experiment will make it impossible for you to ever get laid unless you become a billionaire, so carefully weight what you are ready to give up for science.