Something which was not mentioned in this discussion is the way some winding systems (Pellaton, Magic Lever and everything along that developmental continuum) have variable winding efficiency depending on a combination of direction and position of the rotor. This can be 'tuned' (simply by the adjusting the position of the rotor at the time of assembly) to deliver maximum efficiency in a particular arc e.g. when the crown is pointing generally downwards (i.e. the wearer is walking). Another instance of the difference between theoretical efficiency and 'real world' efficiency.