I get all that. Congestion still goes down. Whenever two google-cars pair up and become (effectively) one longer car, it's a little less congestion. You might argue that it's not enough, but it's surely something. Those google-cars that see you coming on to the 405 - are you assuming they will give you more room than humans would? Because they are responsible drivers? I suppose that might negate some of the congestion gains, but I'm hard-pressed to see any other way congestion could be anything but improved with google-cars.
I would argue that it's an exponential growth curve and the tail is going to be very long. I would further argue that as more people get used to not having to drive, the more things are going to suck when they're forced to in a rainstorm. To be clear: I don't think driverless cars are a bad idea. I do think that their benefits are oversold, their timeline is overly optimistic, and as panaceas for reducing traffic they suck ass compared to viable mass transit.You might argue that it's not enough, but it's surely something.
Fair enough, and agreed on all points.
I think it would be a very bad idea to have driverless cars that people can take over from and drive, ever, for just those reasons you mention. Autonomous vehicles ought not to allow a human driver; just my opinion. "Smart cruise control" - no thanks. Recipe for disaster.