Following up a bit --- I understand the mechanics of Di2 are different from Zap or Mektronic. Mavic had to comply with strict rules regarding what was allowed in terms of rider shift actions. No external power source was allowed to provide any work (i.e. work=force*distance) on the bike. Nothing could be moved unless powered by the human riding the bike. This included front & rear derailleur shift motions. Mavic's system is extremely ingenious (at least regarding the rear derailleur) in that the movement of the pulleys from one cog to another is actually accomplished through the action of internal cams driven by the power of the rider, through harnessing the rotation of the idler pulleys. A really neat design, which is a topic for another post altogether.
In contrast, Shimano's solution of electric motors driving both derailleurs is not too challenging.
There must have been a rule change in this area since the Zap/Mektronic days, which allows the use of non-human power to move shift elements. This explains, to me anyway, why Mavic couldn't solve the front derailleur issue, and always had that manually-powered. Using a motor to move it is simple & basic, but doing it under the old rules -- well, that is a tuffy!
So, I'm guessing Di2 wouldn't be susceptible to cold weather issues a la Mektronic, because Shimano uses a motor to move the cage, and it's driven by (relatively) high-current battery-supplied electricity.
The only issue in this regard that Shimano would have had to design for is allowing for the battery & electrical system performance in cold weather. I just wonder what the low design limit is for this system?
final note: It is promising seeing the photo on the back of the current VeloNews - Nys winning a cold cross race using Di2 !
