When the servo pulls the cable connected directly to the torque rod arm, the arm moves forward and the elevator deflects downwards. When the servo pulls the cable that goes around the pulley then to the torque rod arm, the arm moves backwards and the elevator deflects upwards.

The main reason why I'm thinking about this is for weight savings. Every ounce counts, right? The pushrod alternative requires a 34" length of (aluminum?) pushrod, guided to the tail with a 24" lengh of pastic tube. Although I haven't done any weight comparisons yet, I would think that the weight of two 34" lengths of pull-pull cable plus the weight of the pulley would be much less that the weight of the pushrod assembly. If not, then the pushrod method is a much safer and proven way to go.