And what about the risk to other road users, pedestrians and cyclists, who are not encased in steel crumple zones?
Regulating driving behaviour is not just about preservation of the driver, it's also about creating an environment conducive to a functioning community. At present the bias is strongly towards the car-user ( e.g. making pedestrians wait at crossings ) but at least the principle is maintained.
Even with highway fatalities declining it remains one of the most likely ways to die for younger age groups.
There's a bit of a "paradox" in that likelihood of death per mile driven is extraordinarily low, likelihood of your dying this year from a motor vehicle accident is very low... But if you die this year, the likelihood you died in a car accident, rather than from something else entirely, ends up somewhat high.
In the end I support data-driven lawmaking. We know that speed limits are ignored by 90% of everyone, and we know very well how road design parameters such as lane width affect the way people drive. I think we should ditch speed limits altogether in the city and focus on these things which work.
The actual risk of driving a vehicle is astoundingly low given the physics involved. Wrecks are somewhat common but injuries relative rare and death is such a remote possibility that most people can disregard it compared to all the other ways the universe is out to get them :-).