On January 23 around 7:40 a.m. near Grant Elementary School, Santa Monica Waymo’s driverless Jaguar I-Pace hit a child who ran from behind a double-parked SUV amid crossing guards and stopped cars. The vehicle braked from 17 mph to under 6 mph before minor-impact contact, then called 911 and stayed until police cleared it; the child walked away with minor injuries. NHTSA opened probe PE26001 to check caution near schools, speed compliance, and response protocols, amid prior Waymo issues like school bus failures in Atlanta and Austin.

My analysis shows Waymo is safer than the average US human driver. But the safety of US human drivers is less as I was showing. The range of human driving is broad. A lot of too drunk, too sleepy, too distracted, too old or new drivers mess up the reality. Waymo has systemic issues — getting confused about what to do and then locking up. Brittleness, and depending upon hypermapping. The speed of decisions I expect to be a problem especially in cases of a lot more people and vehicles around in a chaotic situations. Waymo has 10-20X more problem situations beyond the crash risk.

Waymo and Tesla will be or are safe enough and no fundamental limit against become 10-40 times safer than human.

The main Tesla advantages are being able to get to true scale and quickly. Next 3 years Waymo is very limited by the amount of robotaxi grade Lidar. Not enough factories and factory capacity and having to take the sensor kits and then mostly hand build the modifications. 3-5 per day is the current limit they have shown. Tesla pulling the safety monitors and chase cars means austin will become completely unsupervised and then SF Bay Area and 10-20 other metro areas this year with hundreds of thousands of cars and potentially millions with airbnb business model. Any remote monitoring is meaningless because that is the standard and is economically scalable at 3 cars to one remote monitor. 1.1 million with active FSD this could be 2-4 million by year end