Gosh, it's amazing the biases we have when the data is non-evaluatable, and then even when we can compare, we still have a bias towards an overflowing but smaller can, and an under filled but bigger can. The funny thing is I realize I've thought this way, too, until I read this just now. I shall not make the mistake again!
I agree with your idea. It reminds me of your lecture in Stanford: all humans dying is bad because in effect we lose the entire known universe. Someone risking their only chance at sentient existence is much braver than someone who thinks they will live on in heaven. Very cool. Just told my friend and he agreed.