Pundit Accountability

I will say that in 2025, it is clear that "robotaxis" work better than I imagined a few years ago. Still it is also clear that there is an immense amount of human intervention required - from remote operation to gig workers closing the fucking doors - so the dream of a self-driving car world is still a few Friedman Units away.

I am pretty sure that my belief that Waymo continues to obscure a lot of its necessary human intervention behind carefully defined stats is correct.

Though less sure than I once was, I still doubt that Waymo and similar will be anything but incredibly expensive research projects anytime soon. They might continue to improve, but they won't lead to a profitable business model and we are a long way off from true personal robocars of the "take a nap while you drive" kind.

It is a weird business model to pursue. Drivers are cheap and the uber model makes them responsible for their own capital (car) expenses.  Exploitive but hard to beat!