Using wiki numbers, a loaded 787 at maximum range will do some 4.2M person-km on 101 tons of fuel. That's ~0.024kg/km of consumed fuel. A Prius with a single occupant will only do ~0.036kg/km. The aircraft is actually a third more efficient than the car. Now, you can carpool, quartering the Prius's number. Meanwhile, aircraft are not always full (increasingly rare these days), shorter flights are less efficient, regional jets are less efficient, the hub-and-spoke means you're not flying directly to your destination. On the other hand, typical car travel is single occupant, dominated by trucks, SUVs, and large cars with less than half the average rated economy of that Prius, and you're not traveling straight line in a car either.
Where did you come up with three orders of magnitude discrepancy? I could maybe see one vehicle-mile equating to 1500 in a car, which puts passenger-mile production around an order of magnitude higher.
Good for backhauls
Nearly all long distance links are single-mode fiber, because multi-mode fiber results in dispersion and low signal quality at longer distances. You can't run QAM over single-mode fiber. Assuming the breakthrough here is a new modulation technique that allows for such longer distance links, you would still need to run all new fiber.
Electric motors produce motion without losing any mass
Rockets produce motion without losing mass either. The body moves in one direction, the propellant moves in the opposite direction, and the combined system encompassing both remains in place. That is exactly how an electric motor produces motion, because the system as a whole doesn't move.
so I don't really see why it's impossible for there to be some way of producing thrust in a vacuum using only energy.
It's not impossible. You're describing what's called a "photon drive". You shine light out the back, and you have an ever so slight "pressure" that drives you forward. Alternatively, someone else shines a light on your back, and that same pressure drives you forward. This is how solar sails operate. These work, we've tested them experimentally and the results match up with the theory. You just need obscene amounts of power for negligible thrust.
The problem here is now we're talking about something that is several orders of magnitude more efficient in converting energy into momentum.
I expect it's not worth the trouble to implement since any way you do it you need a bunch of extra equipment on the car to figure out what the speed limit is, which makes the feature more expensive
The car already has cameras and GPS, and uses both machine vision and mapping to determine the local speed limit. As of the beginning of this year, speed limits are enforced for residential and undivided roads. US27 is a divided highway, and thus the software only warns that you're speeding.
Maybe they need a way to keep the driver involved: steering wheel pressure sensors, or eye sensors.
The system already does this, and will throw visible and audible warnings, before eventually slowing to a stop if the driver does not take action to confirm they are still maintaining control.
As designed, Tesla's Autopilot is not what many people think an autopilot is; it's a driver-assist feature and not an autonomous driving system.
But Tesla's Autopilot is what an autopilot actually is. It's a driver aid, that exists to reduce a driver's workload while behind the wheel, not eliminate it entirely. Just because a pilot turns the autopilot on does not mean they are magically no longer the pilot and can zone out.
As such it makes people less safe on the road for themselves and others when they engage it;
People make themselves less safe on the road by not understand how their vehicle operates.
1 1 was a race-horse, 2 2 was 1 2. When 1 1 1 1 race, 2 2 1 1 2.