We have seen that Waymo can go haywire. However unlike Tesla, we don't see it broadcasting it on CNN/CNBC/Biden's twitter feed/etc etc. The Waymo's F up youtube video has 420k views with only one local channel picking it up.
This is a huge risk going forward and I see Robotaxi being a thing after a good portion of the population have used Tesla FSD production code and are comfortable with it. That's the only way to battle the unknown before the media shape their own narrative.
To get regulatory approval in most places it will need to be substantially better than a human and have stats to back that up.
It may need to pass specially devised tests, with some random variation.
IMO the ideal situation is what I would call a "Bedrock Version", that is a version that gets most things right, in most situations, and rarely has a regression when retraining., Get that "Bedrock Version", and Tesla can safely retrain to eliminate progressively more and more obscure edge cases.
It will need to be a lot better than what Waymo current has, and signicantly better than the current FSD beta version.
We are all well used to the media beating up small aspects like staff leaving Tesla, car fires, recalls autopilot crashes, often these have share price impact. But mostly that impact should be retracing some of the gains when FSD was complete, or when FSD was approved. So any drop is from a higher price than we have today.
The maximum I expect regulators to do is, suspend Tesla FSD operations in an area while any incident is investigated, that will grind through a process, and after sometime I expect the FSD licence would be reinstated.
Is bad press anything new or unexpected?
Overtime the stats should say on average signicantly safer than human that is all that counts. Some humans are dragging the human average down, in terms of building the case for FSD, that is a big help.
I 'll take self driving any day.
