Autonomous Tesla Delivery [video]
youtube.com
youtube.com
As a example, by end of 2016 Waymo was doing 5,000 miles between interventions [1]. It took them until end of 2020 to reach ~30,000 miles per intervention [2]. They only began driverless operations in 2022. 6 years of improvement from a capability level of 5,000 miles per intervention before they thought it was safe enough to hit the roads with no driver.
In contrast, Tesla is averaging 500 miles at best [3] and the incident rate of their pilot program corroborates that their "Tesla Robotaxi" version is not materially better. As such, getting a cherry picked 15 mile journey should be expected, but is not at all indicative of actual adequacy for scalable autonomous operation. And despite that, they still failed to actually follow the law for the entire duration failing on a task as trivial as not parking in a gigantic red fire lane.
This is also in spite of starting their autonomous driving development by 2016 at the latest. So, after 9 grueling years they are still too immature to handle simple signage such as fire lanes. After 9 grueling years they have achieved much less than what Waymo had in 2016, the year they started. If Tesla literally time-traveled their current technology back to when they started, it would be 10 times worse than cutting edge.
Furthermore, Waymo was founded in 2009. It took Tesla 9 years to achieve significantly less than what Waymo achieved in 7 years, despite Waymo starting 7 years earlier. So Tesla's rate of improvement is slower and they still have not yet caught up to where Waymo was when Tesla began their program. And Waymo thought they needed 6 years of further development beyond that level of performance before it was not a unnecessary risk to the public to go fully driverless.
Pointing at single drives as evidence of safety and capability is nonsense. A single drive provides 1/3,000th the evidence you need to conclude that they can go 3,000 drives in a row without problems. But breaking the law on your first drive provides 2,999/3,000th the evidence you need to conclude that they can not go 3,000 drives in a row without problems.
[1] https://thelastdriverlicenseholder.com/2018/02/01/disengagem...
[2] https://thelastdriverlicenseholder.com/2021/02/09/2020-disen...
It is unfortunate that there is no higher quality source of data as the manufacturer deliberately hides their data and only publishes deliberately deceptive, juvenile, and unscientific reports that disregard even basic scientific requirements you learn to apply in high school such as error bars.
If you care to present a robust, representative, auditable, high-sample dataset supporting your apparent assertion that the numbers are dramatically higher, be my guest. And no, arguing that there is no available high quality data thus you get to claim whatever number or safety you want without evidence is not a valid argument.
That the manufacturer is unable to present any scientifically sound data to support their assertions, and in fact only producing deliberately deceptive conclusions, despite literal billions of miles, billions of dollars, and years of work is enough to switch the burden of proof onto them.