> Sandy generally has great commentary on the auto industry. He's quite genuine
I'm sure he does (I say that sincerely). But he's also the same Sandy who was very anti-Tesla until his wife got one and he put a whole lot of his investment portfolio in TSLA stock.
> There are many drivers on the road, many of who make the road unsafer.
Many of those drivers may make poor choices, etc. But we're not touting them as role models or desired behavior. FSD is meant to be better, and is still _actively_ trying to do things that are dangerous to drivers and pedestrians.
> If FSD beta is better on average, then the roads are getting stepwise safer.
The sheer amount of stupid decisions I've seen from FSD on YouTube etc. is shocking. The only thing more stupefying is watching people actively "allowing" it to do stupid things for "views", their own inattentiveness, or for some misguided notion that they're somehow "contributing" to "training" the car.
Like pro tip: When your car has run a red light at the same intersection the last three times, don't let it do it a fourth time just "to help the project". That's irresponsible, reckless and outright dangerous.
Tesla flat out refuses to release any statistics that would actually be able to prove, or disprove, an actual fair comparison of the safety of FSD/AP. Instead we get these cherry picked miles driven where AP is touted as being safer - on the (unspoken) subset of miles where it didn't disengage because it couldn't drive or the driver decided it was safer to disengage because of the AP's driving, versus "human drivers, all miles, all conditions".
There's (relatively) few accidents for human drivers on well-painted, well-maintained roads in the California sunshine. And similar for AP/FSD.
There's a lot more accidents for human drivers on poorly painted and maintained roads in a Pittsburgh winter. And there would be for AP/FSD too, if it wasn't disengaged a large majority of that driving.
But Tesla happily lumps the sum of driving of "Human drivers in California sun and Pittsburgh winter" and says that mile-for-mile, AP/FSD is better because it had fewer accidents, even though most of those miles were accumulated in the California sun. Anyone who completed high school statistics would know this wasn't a fair comparison, but Tesla has pushed it for years.