There are plenty of plausible explanations for those statistics where the comparison isn't a good one. The 479K figure given appears to be an average over the entire population of vehicles and drivers.

But what is the performance like for other drivers of expensive cars? Teslas aren't cheap, so it's a reasonable hypothesis that their drivers might be more likely to take care of them than the owners of a 20-year-old hand-me-down that is nearing the end of its useful lifetime anyway.

What is the performance like for drivers in other cars that have relatively good driver aid technologies? This might be correlated with more expensive cars, but more basic models typically gain a lot of the safety tech as well after a while.

What is the performance like for other drivers of the same ages and experience levels as Tesla drivers? Given the cost of the vehicle, Tesla drivers might tend to be older and more experienced than average, and both of those factors are known to be strongly correlated with safer driving.

Or maybe Tesla's own statistics are based on such a small population that they might not be representative anyway?

Or maybe Tesla's "registration" of accidents is flawed, and lots of minor incidents go unreported in Tesla's stats but some of them would be counted by NHTSA?

Or maybe Tesla's touchscreen controls are simply so complicated or annoying that many drivers avoid using them as much as possible?

As I said, there are many plausible explanations that would make that "amazing real world safety data" look a lot less impressive. It's impossible to evaluate that record intelligently based on just the data you quoted.

Don't you think the data from Tesla's computers on wheels could be more accurate than the second hand data (accident occurs, is written down, then provided to the government agency) used by the NHTSA and other researchers. Surely there is some loss here in unreported incidents that Tesla would not lose, making the NHTSA reports actually better than the reality potentially.

Could be? Sure.

Definitely is? Again, there's no way to tell from just the data you cited.

We don't know what Tesla counts as "registering an accident", and clearly it isn't a neutral observer here.

20% of vehicle crashes in the US are caused by motorcycles, and the 479kmi number includes those. The statistics have always been misleading for Tesla to use those numbers to compare vehicle safety. When comparing just luxury cars, Teslas have far more fatalities than other cars in the same price range.

Important to note that this data on the fatality rate was put together by a known short seller and is from before the model 3 was released which has the lowest probably of injury of any vehicle ever tested by the NHTSA. Also before many improvements have been made to both the model s, 3 and x software and hardware over the years.

https://www.tesla.com/blog/model-3-lowest-probability-injury...

Do you have a source about the motorcycles being included?

If I was asked to explain it, I would start by researching whether people who own Teslas might be safer drivers than average when they weren't in Teslas. Or whether, by virtue of other characteristics of the car, like handling, acceleration, and breaking, Teslas may be safer in general, regardless of their dashboards. I'd also question the source of the data and the methodology by which it was collected, just to be diligent. I wouldn't just start by assuming that having a touch screen made cars safer, though.