Interesting idea. Half a dozen choices is nowhere near enough--when I tried setting up my own grid, I had a tough time getting below eight--and some pairs were close enough that it was nearly a toss-up which one I ended up picking. But half a dozen choices times twenty or thirty is a lot less binary, and making people prioritize that sharply means that common unexpected results probably do have real meaning.

It would be interesting to compare the results of this test with one that's a bit more nuanced--rank each trait from 1 to 5, maybe, and then see which characteristics your friends ranked higher than you, and which lower.