My study of electricity and magnetism is now about 55 years old, so I am not going to claim I fully understand the pros and cons of shorting the alternator coils. Therefore take the following with some caution.
I think shorting the coils does more than simply directing current to ground (earth), with a commensurate heat gain. There is also the effect on the magnetic flux within the stator. Stators are made with laminated iron cores for a reason that I do not remember, but one can expect the iron core improves the generation of current, otherwise it would not be there.
Now consider the older three wire stators: the sketch below shows how these were internally wired.
Before the days of Zener diodes and electronic regulators, the main lighting switch controlled the alternator's output to match the load.
In the OFF position, the lighting switch connected GY to WG, i.e. shorting the parallel pair of coils. The battery was being supplied by the GB series pair, and the current delivery to the load was MINIMAL in this configuration.
In the PILOT position, the lighting switch open circuited GY, and the current output of the alternator was increased.
Here is the crux of the issue: the same series pair of coils, connected to the GB wire, increased the current output, when the shorted coils were open circuited. This indicates shorting the coils affected the magnetic flux in some way, which IIRC from Physics 101, affected the flux in the iron core.
In the HEAD position, the lighting switch connected GY to GB, giving full alternator output.
The function of the lighting switch can be verified by studying the switch positions in the wiring diagram given for the 650 American (the 7th down from the top):
https://www.accessnorton.com/NortonCommando/early-norton-wiring-diagrams.20232/
The bottom line: shorting the coils affects the magnetic flux and decreases the stator output. Now (take this with caution), might not shorting the coils in a 2 wire stator, similarly affect the magnetic flux and decrease the induced current in the coils as well, thereby not creating the heat load one expects? Wish I could positively answer that, but I think it is worth considering.
Slick