The whole argument in favor of a g factor in intelligence is that when you measure different kinds of intelligence (verbal, analytical etc.) the numbers you get seem to be a function of a single dominant factor. (That's the argument, I'm not saying whether or not I agree with it.)
A user could sort their view of the front page based on either props (agree votes) or recommendations (interest votes) or some weighted combination. Maybe too complicated?
But I've concluded that I really don't understand people that well, so I don't have a good intuition as to what would actually work.
False analogy. Karma-score blending is arithmetic. Latent factors of mental-ability tests (or latent-factors of anything) are derived non-arithmetically.