Far-term horizon. Despite big claims about driverless car tech, it's not reliable enough for production.

The thing is that, at the tipping point, they become safer than human drivers (sensors) which means 1) less accidents and 2) less traffic (orderly flow, no butterfly effect drivers). In addition they'll self park or be in constant motion (public transportation) making the a more efficient than regular cars. So you'll see government pressure and industry ($500 more per year if you want to 'self drive' for registration and insurance) There'll be dedicated SDE lanes that'll run smoother and squeeze self drivers in with human error prone and erratic peers. That tipping point will get us there very quickly from say Year 7 to Year 10 in major urban markets.

Current tech doesn't work in rain or snow. Computer vision cannot tell if a car is moving behind a bush, or if it's a bird or a plastic bag. It doesn't work off-road. These issues, and many others, will take a lot more than 10 years to fix. There is also no guarantee that Google will profit from driverless cars when they do arrive. The whole driverless car project is simply a way for Google to distract investors, the press, and the public from the lack of ideas in their core search and advertising business.

And even if it were, worldwide adoption will be measured in decades if it follows the current automobile aging curve

Right but I don't think it follows traditional curves because this isn't an incremental change in automobiles this is a much broader change with bigger social benefits (Billions saved in lost and injured lives/faster delivery times/better quality of life)