The biggest problem with this article is that it takes as a given that commercial autonomous taxi services are inevitable. The data being collected here include

The biggest problem with this article is that it takes as a given that commercial autonomous taxi services are inevitable. The data being collected here includes only trips in a pristinely mapped area. There’s no data about how often service is suspended because of weather, or how much labor beyond the safety driver is required to keep the vehicles in operational order, or how sensitive the sensors are to common visibility issues that happen in the real world, or which rides are turned down because of construction or other temporal factors, or what the failure states would be without a human safety driver... But even if Waymo beats all the odds and common sense and makes an autonomous vehicle that can function as a taxi in a way people actually want to use outside of highly regularized low-density suburban environment, I continue to be surprised that anyone thinks these services would be able to operate in a profitable way, or that they will change anything about how cities work except for doubling automobile traffic and irritating any human drivers, cyclists, and pedestrians that are forced to interact with them.