If your system kills someone every 10 million miles (10x worse than US average) and you only sell 100 of them it’s unlikely it will ever kill anyone considering it will only be used for a small fraction of vehicle miles.
I was beat to it, but it's all a game of probabilities. If you only have a 100 cars, probability says you are far less likely to have a single accident. If you have a million cars using this system, the probability you get a single accident (or more) is much higher. Lease-only also gives you a mechanism to limit miles driven (most leases have mileage limits).
Yeah - probabilities only work until something happens. Then the probability is 100%. This approach seems a bit like a cop out. I think Tesla's approach of starting out with a reasonably wide release of beta testers to get data, tweak and fix bugs until you know you have a good system is better, or do like Mercedes do - release it with very limited abilities and use cases.
I don't think it's a cop out, but rather how the auto industry has done things since probably forever to calculate accepted risk.
The first radar cruise control was also available only on the S-class. Tesla aiming to releaseL5 capability on even their lowest price model is very unusual for a new automotive technology. I can’t think of anything analogous.
Appears you are unfamiliar with the calculations used by large companies in the practice of risk management and the application of actuarial science
Well, on their lowest price model for an additional $12,000 upcharge.
Requiring driver supervision is the ultimate cop out when it comes to full self-driving.
There are literally robotaxis on the road right now with nobody in the driver's seat. I agree that it's going to be a long time before they cover a significant percentage of roads.
Comparing the lack of drivers in robotaxis to the lack of driverless trains or driverless aircraft is not really appropriate, not because of similarities or differences in the operational design domains (ODDs), but because of the issue of incremental "cost" of having a driver per passenger-mile traveled. Having worked with autopilot systems in aircraft, IMO it would be FAR simpler to automate and remote-control trains and planes, and the safety margins are marginally good, but the costs of, e.g., having two pilots in a 250-passenger plane are just not as significant as, say, fuel. In a taxi, most of the cost is the driver.
I understand that...in fact most of the problem is union power...protecting jobs. And you can see the action that taxi drivers around the world have taken against Uber to know that they won’t take Robotaxis lying down either
To me it looks like this was a miscalculation: They believed that reaching L5 with vision and AP/HW2 would have been much easier. If that would have been the case, it would have made a lot of sense.
Absolutely the opposite.
I generally agree but I think there would be some benefit for Tesla to be able to say they have true level 3 on highways. I think you're right, though - the financial liability is still a significant concern for investors.
