Interesting, I'm curious why this analogy was chosen, and not the other way around. Current as torque and voltage as (rotational) velocity would seem to allow the junctions to be much simpler -- just a gang of gears on a single axle, rather than the differential mechanism used here -- since torques (like current) add at an axle/junction, while rotational velocity (like voltage) transmits through an axle/junction.

Edit: I suppose a/the major reason is to get the resistance = friction analogy down. With current as torque, friction acts as conductance. E.g., a constant-torque/"current" motor would need to be held still (high friction) to prevent its speed/"voltage" from growing, whereas with the equivalent electrical circuit, you need to short the terminals (low resistance) of a current source to do the same.

I suppose the reason for this seeming discrepancy is that, in an electric circuit, wires are separated by high resistance. But in a physical circuit, "wires" are separated by low friction (= not touching). Flipping around the natural behavior of a junction allows you to take the dual of the entire circuit, thus causing the behavior of resistance and friction to line up.